This is an archive of the discontinued LLVM Phabricator instance.

Differential D136903

[OpenMP][NFCI] Remove effectively dead code from clang and the runtime
ClosedPublic

Authored by jdoerfert on Oct 27 2022, 5:01 PM.

Details

Reviewers
jhuber6
tianshilei1992
ABataev
jsjodin
Commits
rG90609fb68f05: [OpenMP][NFCI] Remove effectively dead code in clang and the runtime

Diff Detail

Event Timeline

jdoerfert created this revision.Oct 27 2022, 5:01 PM
Herald added a project: Restricted Project. · View Herald TranscriptOct 27 2022, 5:01 PM
Herald added subscribers: guansong, bollu, hiraditya, yaxunl. · View Herald Transcript
jdoerfert requested review of this revision.Oct 27 2022, 5:01 PM
Herald added a project: Restricted Project. · View Herald TranscriptOct 27 2022, 5:01 PM
Herald added subscribers: llvm-commits, sstefan1. · View Herald Transcript
jdoerfert added a comment.Oct 27 2022, 5:01 PM

I'll fix the tests up soon, this is a heads up.

Harbormaster completed remote builds in B194786: Diff 471333.Oct 27 2022, 5:02 PM
jdoerfert updated this revision to Diff 481461.Dec 8 2022, 3:20 PM
Herald added subscribers: kosarev, mattd, asavonic and 3 others. · View Herald TranscriptDec 8 2022, 3:20 PM
jhuber6 accepted this revision.Dec 8 2022, 3:23 PM

Nice, about time we got rid of that unused argument.

This revision is now accepted and ready to land.Dec 8 2022, 3:23 PM
Harbormaster completed remote builds in B202097: Diff 481461.Dec 8 2022, 8:13 PM
jdoerfert added a comment.Dec 12 2022, 10:09 PM

Committed as c1c8cbbf5f29257d084a23a2f6c4236c40b7afb9, one test fails apparently non-deterministically, reverted in f9c29878b04b2d0725eb05716363b27bcda22a5f.

This revision was landed with ongoing or failed builds.Dec 13 2022, 7:01 PM
Closed by commit rG90609fb68f05: [OpenMP][NFCI] Remove effectively dead code in clang and the runtime (authored by jdoerfert). · Explain Why
This revision was automatically updated to reflect the committed changes.
jdoerfert added a commit: rG90609fb68f05: [OpenMP][NFCI] Remove effectively dead code in clang and the runtime.
Herald added projects: Restricted Project, Restricted Project. · View Herald TranscriptDec 13 2022, 7:01 PM
Herald added subscribers: openmp-commits, cfe-commits. · View Herald Transcript

Revision Contents

PathSize
clang/
lib/
CodeGen/
16 lines
52 lines
test/
OpenMP/
4 lines
14 lines
128 lines
24 lines
358 lines
1032 lines
32 lines
40 lines
88 lines
20 lines
298 lines
48 lines
48 lines
12 lines
8 lines
24 lines
16 lines
52 lines
24 lines
144 lines
16 lines
64 lines
16 lines
48 lines
3012 lines
4 lines
2 lines
2 lines
590 lines
84 lines
144 lines
404 lines
370 lines
634 lines
682 lines
124 lines
12 lines
520 lines
694 lines
1134 lines
126 lines
802 lines
370 lines
950 lines
572 lines
2200 lines
llvm/
include/
llvm/
Frontend/
OpenMP/
8 lines
4 lines
lib/
Frontend/
OpenMP/
14 lines
Transforms/
IPO/
21 lines
test/
Transforms/
OpenMP/
12 lines
1 line
184 lines
100 lines
12 lines
12 lines
16 lines
28 lines
28 lines
27 lines
8 lines
8 lines
36 lines
28 lines
32 lines
36 lines
8 lines
140 lines
12 lines
11 lines
24 lines
36 lines
12 lines
openmp/
libomptarget/
DeviceRTL/
include/
4 lines
src/
4 lines

Diff 482688

clang/lib/CodeGen/CGOpenMPRuntimeGPU.h

Show First 20 Lines • Show All 147 Lines • ▼ Show 20 Linesprotected:
StringRef getOutlinedHelperName() const override { StringRef getOutlinedHelperName() const override {
return "__omp_outlined__"; return "__omp_outlined__";
} }
/// Check if the default location must be constant. /// Check if the default location must be constant.
/// Constant for NVPTX for better optimization. /// Constant for NVPTX for better optimization.
bool isDefaultLocationConstant() const override { return true; } bool isDefaultLocationConstant() const override { return true; }
/// Returns additional flags that can be stored in reserved_2 field of the
/// default location.
/// For NVPTX target contains data about SPMD/Non-SPMD execution mode +
/// Full/Lightweight runtime mode. Used for better optimization.
unsigned getDefaultLocationReserved2Flags() const override;
public: public:
explicit CGOpenMPRuntimeGPU(CodeGenModule &CGM); explicit CGOpenMPRuntimeGPU(CodeGenModule &CGM);
void clear() override; void clear() override;
bool isTargetCodegen() const override { return true; }; bool isTargetCodegen() const override { return true; };
/// Declare generalized virtual functions which need to be defined /// Declare generalized virtual functions which need to be defined
/// by all specializations of OpenMPGPURuntime Targets like AMDGCN /// by all specializations of OpenMPGPURuntime Targets like AMDGCN
▲ Show 20 Lines • Show All 205 Lines • ▼ Show 20 Lines
private: private:
/// Track the execution mode when codegening directives within a target /// Track the execution mode when codegening directives within a target
/// region. The appropriate mode (SPMD/NON-SPMD) is set on entry to the /// region. The appropriate mode (SPMD/NON-SPMD) is set on entry to the
/// target region and used by containing directives such as 'parallel' /// target region and used by containing directives such as 'parallel'
/// to emit optimized code. /// to emit optimized code.
ExecutionMode CurrentExecutionMode = EM_Unknown; ExecutionMode CurrentExecutionMode = EM_Unknown;
/// true if we're emitting the code for the target region and next parallel
/// region is L0 for sure.
bool IsInTargetMasterThreadRegion = false;
/// true if currently emitting code for target/teams/distribute region, false /// true if currently emitting code for target/teams/distribute region, false
/// - otherwise. /// - otherwise.
bool IsInTTDRegion = false; bool IsInTTDRegion = false;
/// true if we're definitely in the parallel region.
bool IsInParallelRegion = false;
/// Map between an outlined function and its wrapper. /// Map between an outlined function and its wrapper.
llvm::DenseMap<llvm::Function *, llvm::Function *> WrapperFunctionsMap; llvm::DenseMap<llvm::Function *, llvm::Function *> WrapperFunctionsMap;
/// Emit function which wraps the outline parallel region /// Emit function which wraps the outline parallel region
/// and controls the parameters which are passed to this function. /// and controls the parameters which are passed to this function.
/// The wrapper ensures that the outlined function is called /// The wrapper ensures that the outlined function is called
/// with the correct arguments when data is shared. /// with the correct arguments when data is shared.
llvm::Function *createParallelDataSharingWrapper( llvm::Function *createParallelDataSharingWrapper(
llvm::Function *OutlinedParallelFn, const OMPExecutableDirective &D); llvm::Function *OutlinedParallelFn, const OMPExecutableDirective &D);
/// The data for the single globalized variable. /// The data for the single globalized variable.
struct MappedVarData { struct MappedVarData {
/// Corresponding field in the global record. /// Corresponding field in the global record.
llvm::Value *GlobalizedVal = nullptr; llvm::Value *GlobalizedVal = nullptr;
/// Corresponding address. /// Corresponding address.
Address PrivateAddr = Address::invalid(); Address PrivateAddr = Address::invalid();
}; };
/// The map of local variables to their addresses in the global memory. /// The map of local variables to their addresses in the global memory.
using DeclToAddrMapTy = llvm::MapVector<const Decl *, MappedVarData>; using DeclToAddrMapTy = llvm::MapVector<const Decl *, MappedVarData>;
/// Set of the parameters passed by value escaping OpenMP context. /// Set of the parameters passed by value escaping OpenMP context.
using EscapedParamsTy = llvm::SmallPtrSet<const Decl *, 4>; using EscapedParamsTy = llvm::SmallPtrSet<const Decl *, 4>;
struct FunctionData { struct FunctionData {
DeclToAddrMapTy LocalVarData; DeclToAddrMapTy LocalVarData;
llvm::Optional<DeclToAddrMapTy> SecondaryLocalVarData = std::nullopt;
EscapedParamsTy EscapedParameters; EscapedParamsTy EscapedParameters;
llvm::SmallVector<const ValueDecl*, 4> EscapedVariableLengthDecls; llvm::SmallVector<const ValueDecl*, 4> EscapedVariableLengthDecls;
llvm::SmallVector<std::pair<llvm::Value *, llvm::Value *>, 4> llvm::SmallVector<std::pair<llvm::Value *, llvm::Value *>, 4>
EscapedVariableLengthDeclsAddrs; EscapedVariableLengthDeclsAddrs;
llvm::Value *IsInSPMDModeFlag = nullptr;
std::unique_ptr<CodeGenFunction::OMPMapVars> MappedParams; std::unique_ptr<CodeGenFunction::OMPMapVars> MappedParams;
}; };
/// Maps the function to the list of the globalized variables with their /// Maps the function to the list of the globalized variables with their
/// addresses. /// addresses.
llvm::SmallDenseMap<llvm::Function *, FunctionData> FunctionGlobalizedDecls; llvm::SmallDenseMap<llvm::Function *, FunctionData> FunctionGlobalizedDecls;
llvm::GlobalVariable *KernelTeamsReductionPtr = nullptr; llvm::GlobalVariable *KernelTeamsReductionPtr = nullptr;
/// List of the records with the list of fields for the reductions across the /// List of the records with the list of fields for the reductions across the
/// teams. Used to build the intermediate buffer for the fast teams /// teams. Used to build the intermediate buffer for the fast teams
/// reductions. /// reductions.
/// All the records are gathered into a union `union.type` is created. /// All the records are gathered into a union `union.type` is created.
llvm::SmallVector<const RecordDecl *, 4> TeamsReductions; llvm::SmallVector<const RecordDecl *, 4> TeamsReductions;
/// Shared pointer for the global memory in the global memory buffer used for
/// the given kernel.
llvm::GlobalVariable *KernelStaticGlobalized = nullptr;
/// Pair of the Non-SPMD team and all reductions variables in this team /// Pair of the Non-SPMD team and all reductions variables in this team
/// region. /// region.
std::pair<const Decl *, llvm::SmallVector<const ValueDecl *, 4>> std::pair<const Decl *, llvm::SmallVector<const ValueDecl *, 4>>
TeamAndReductions; TeamAndReductions;
}; };
} // CodeGen namespace. } // CodeGen namespace.
} // clang namespace. } // clang namespace.
#endif // LLVM_CLANG_LIB_CODEGEN_CGOPENMPRUNTIMEGPU_H #endif // LLVM_CLANG_LIB_CODEGEN_CGOPENMPRUNTIMEGPU_H

clang/lib/CodeGen/CGOpenMPRuntimeGPU.cpp

Show First 20 Lines • Show All 88 Lines • ▼ Show 20 Lines
/// code. For all practical purposes this is fine because the configuration /// code. For all practical purposes this is fine because the configuration
/// is the same for all known NVPTX architectures. /// is the same for all known NVPTX architectures.
enum MachineConfiguration : unsigned { enum MachineConfiguration : unsigned {
/// See "llvm/Frontend/OpenMP/OMPGridValues.h" for various related target /// See "llvm/Frontend/OpenMP/OMPGridValues.h" for various related target
/// specific Grid Values like GV_Warp_Size, GV_Slot_Size /// specific Grid Values like GV_Warp_Size, GV_Slot_Size
/// Global memory alignment for performance. /// Global memory alignment for performance.
GlobalMemoryAlignment = 128, GlobalMemoryAlignment = 128,
/// Maximal size of the shared memory buffer.
SharedMemorySize = 128,
}; };
static const ValueDecl *getPrivateItem(const Expr *RefExpr) { static const ValueDecl *getPrivateItem(const Expr *RefExpr) {
RefExpr = RefExpr->IgnoreParens(); RefExpr = RefExpr->IgnoreParens();
if (const auto *ASE = dyn_cast<ArraySubscriptExpr>(RefExpr)) { if (const auto *ASE = dyn_cast<ArraySubscriptExpr>(RefExpr)) {
const Expr *Base = ASE->getBase()->IgnoreParenImpCasts(); const Expr *Base = ASE->getBase()->IgnoreParenImpCasts();
while (const auto *TempASE = dyn_cast<ArraySubscriptExpr>(Base)) while (const auto *TempASE = dyn_cast<ArraySubscriptExpr>(Base))
Base = TempASE->getBase()->IgnoreParenImpCasts(); Base = TempASE->getBase()->IgnoreParenImpCasts();
▲ Show 20 Lines • Show All 658 Lines • ▼ Show 20 Linesvoid CGOpenMPRuntimeGPU::emitNonSPMDKernel(const OMPExecutableDirective &D,
emitTargetOutlinedFunctionHelper(D, ParentName, OutlinedFn, OutlinedFnID, emitTargetOutlinedFunctionHelper(D, ParentName, OutlinedFn, OutlinedFnID,
IsOffloadEntry, CodeGen); IsOffloadEntry, CodeGen);
IsInTTDRegion = false; IsInTTDRegion = false;
} }
void CGOpenMPRuntimeGPU::emitKernelInit(CodeGenFunction &CGF, void CGOpenMPRuntimeGPU::emitKernelInit(CodeGenFunction &CGF,
EntryFunctionState &EST, bool IsSPMD) { EntryFunctionState &EST, bool IsSPMD) {
CGBuilderTy &Bld = CGF.Builder; CGBuilderTy &Bld = CGF.Builder;
Bld.restoreIP(OMPBuilder.createTargetInit(Bld, IsSPMD, true)); Bld.restoreIP(OMPBuilder.createTargetInit(Bld, IsSPMD));
IsInTargetMasterThreadRegion = IsSPMD;
if (!IsSPMD) if (!IsSPMD)
emitGenericVarsProlog(CGF, EST.Loc); emitGenericVarsProlog(CGF, EST.Loc);
} }
void CGOpenMPRuntimeGPU::emitKernelDeinit(CodeGenFunction &CGF, void CGOpenMPRuntimeGPU::emitKernelDeinit(CodeGenFunction &CGF,
EntryFunctionState &EST, EntryFunctionState &EST,
bool IsSPMD) { bool IsSPMD) {
if (!IsSPMD) if (!IsSPMD)
emitGenericVarsEpilog(CGF); emitGenericVarsEpilog(CGF);
CGBuilderTy &Bld = CGF.Builder; CGBuilderTy &Bld = CGF.Builder;
OMPBuilder.createTargetDeinit(Bld, IsSPMD, true); OMPBuilder.createTargetDeinit(Bld, IsSPMD);
} }
void CGOpenMPRuntimeGPU::emitSPMDKernel(const OMPExecutableDirective &D, void CGOpenMPRuntimeGPU::emitSPMDKernel(const OMPExecutableDirective &D,
StringRef ParentName, StringRef ParentName,
llvm::Function *&OutlinedFn, llvm::Function *&OutlinedFn,
llvm::Constant *&OutlinedFnID, llvm::Constant *&OutlinedFnID,
bool IsOffloadEntry, bool IsOffloadEntry,
const RegionCodeGenTy &CodeGen) { const RegionCodeGenTy &CodeGen) {
▲ Show 20 Lines • Show All 75 Lines • ▼ Show 20 Linesenum ModeFlagsTy : unsigned {
LLVM_MARK_AS_BITMASK_ENUM(/*LargestValue=*/KMP_IDENT_SIMPLE_RT_MODE) LLVM_MARK_AS_BITMASK_ENUM(/*LargestValue=*/KMP_IDENT_SIMPLE_RT_MODE)
}; };
/// Special mode Undefined. Is the combination of Non-SPMD mode + SimpleRuntime. /// Special mode Undefined. Is the combination of Non-SPMD mode + SimpleRuntime.
static const ModeFlagsTy UndefinedMode = static const ModeFlagsTy UndefinedMode =
(~KMP_IDENT_SPMD_MODE) & KMP_IDENT_SIMPLE_RT_MODE; (~KMP_IDENT_SPMD_MODE) & KMP_IDENT_SIMPLE_RT_MODE;
} // anonymous namespace } // anonymous namespace
unsigned CGOpenMPRuntimeGPU::getDefaultLocationReserved2Flags() const {
switch (getExecutionMode()) {
case EM_SPMD:
return KMP_IDENT_SPMD_MODE & (~KMP_IDENT_SIMPLE_RT_MODE);
case EM_NonSPMD:
return (~KMP_IDENT_SPMD_MODE) & (~KMP_IDENT_SIMPLE_RT_MODE);
case EM_Unknown:
return UndefinedMode;
}
llvm_unreachable("Unknown flags are requested.");
}
CGOpenMPRuntimeGPU::CGOpenMPRuntimeGPU(CodeGenModule &CGM) CGOpenMPRuntimeGPU::CGOpenMPRuntimeGPU(CodeGenModule &CGM)
: CGOpenMPRuntime(CGM) { : CGOpenMPRuntime(CGM) {
llvm::OpenMPIRBuilderConfig Config(CGM.getLangOpts().OpenMPIsDevice, true, llvm::OpenMPIRBuilderConfig Config(CGM.getLangOpts().OpenMPIsDevice, true,
hasRequiresUnifiedSharedMemory(), hasRequiresUnifiedSharedMemory(),
CGM.getLangOpts().OpenMPOffloadMandatory); CGM.getLangOpts().OpenMPOffloadMandatory);
OMPBuilder.setConfig(Config); OMPBuilder.setConfig(Config);
OffloadEntriesInfoManager.setConfig(Config); OffloadEntriesInfoManager.setConfig(Config);
Show All 36 Linesvoid CGOpenMPRuntimeGPU::emitNumTeamsClause(CodeGenFunction &CGF,
const Expr *NumTeams, const Expr *NumTeams,
const Expr *ThreadLimit, const Expr *ThreadLimit,
SourceLocation Loc) {} SourceLocation Loc) {}
llvm::Function *CGOpenMPRuntimeGPU::emitParallelOutlinedFunction( llvm::Function *CGOpenMPRuntimeGPU::emitParallelOutlinedFunction(
const OMPExecutableDirective &D, const VarDecl *ThreadIDVar, const OMPExecutableDirective &D, const VarDecl *ThreadIDVar,
OpenMPDirectiveKind InnermostKind, const RegionCodeGenTy &CodeGen) { OpenMPDirectiveKind InnermostKind, const RegionCodeGenTy &CodeGen) {
// Emit target region as a standalone region. // Emit target region as a standalone region.
class NVPTXPrePostActionTy : public PrePostActionTy {
bool &IsInParallelRegion;
bool PrevIsInParallelRegion;
public:
NVPTXPrePostActionTy(bool &IsInParallelRegion)
: IsInParallelRegion(IsInParallelRegion) {}
void Enter(CodeGenFunction &CGF) override {
PrevIsInParallelRegion = IsInParallelRegion;
IsInParallelRegion = true;
}
void Exit(CodeGenFunction &CGF) override {
IsInParallelRegion = PrevIsInParallelRegion;
}
} Action(IsInParallelRegion);
CodeGen.setAction(Action);
bool PrevIsInTTDRegion = IsInTTDRegion; bool PrevIsInTTDRegion = IsInTTDRegion;
IsInTTDRegion = false; IsInTTDRegion = false;
bool PrevIsInTargetMasterThreadRegion = IsInTargetMasterThreadRegion;
IsInTargetMasterThreadRegion = false;
auto *OutlinedFun = auto *OutlinedFun =
cast<llvm::Function>(CGOpenMPRuntime::emitParallelOutlinedFunction( cast<llvm::Function>(CGOpenMPRuntime::emitParallelOutlinedFunction(
D, ThreadIDVar, InnermostKind, CodeGen)); D, ThreadIDVar, InnermostKind, CodeGen));
IsInTargetMasterThreadRegion = PrevIsInTargetMasterThreadRegion;
IsInTTDRegion = PrevIsInTTDRegion; IsInTTDRegion = PrevIsInTTDRegion;
if (getExecutionMode() != CGOpenMPRuntimeGPU::EM_SPMD && if (getExecutionMode() != CGOpenMPRuntimeGPU::EM_SPMD) {
!IsInParallelRegion) {
llvm::Function *WrapperFun = llvm::Function *WrapperFun =
createParallelDataSharingWrapper(OutlinedFun, D); createParallelDataSharingWrapper(OutlinedFun, D);
WrapperFunctionsMap[OutlinedFun] = WrapperFun; WrapperFunctionsMap[OutlinedFun] = WrapperFun;
} }
return OutlinedFun; return OutlinedFun;
} }
▲ Show 20 Lines • Show All 2,375 Lines • ▼ Show 20 Lines I->getSecond().EscapedParameters.insert(
VarChecker.getEscapedParameters().end()); VarChecker.getEscapedParameters().end());
I->getSecond().EscapedVariableLengthDecls.append( I->getSecond().EscapedVariableLengthDecls.append(
EscapedVariableLengthDecls.begin(), EscapedVariableLengthDecls.end()); EscapedVariableLengthDecls.begin(), EscapedVariableLengthDecls.end());
DeclToAddrMapTy &Data = I->getSecond().LocalVarData; DeclToAddrMapTy &Data = I->getSecond().LocalVarData;
for (const ValueDecl *VD : VarChecker.getEscapedDecls()) { for (const ValueDecl *VD : VarChecker.getEscapedDecls()) {
assert(VD->isCanonicalDecl() && "Expected canonical declaration"); assert(VD->isCanonicalDecl() && "Expected canonical declaration");
Data.insert(std::make_pair(VD, MappedVarData())); Data.insert(std::make_pair(VD, MappedVarData()));
} }
if (!IsInTTDRegion && !NeedToDelayGlobalization && !IsInParallelRegion) {
CheckVarsEscapingDeclContext VarChecker(CGF, std::nullopt);
VarChecker.Visit(Body);
I->getSecond().SecondaryLocalVarData.emplace();
DeclToAddrMapTy &Data = *I->getSecond().SecondaryLocalVarData;
for (const ValueDecl *VD : VarChecker.getEscapedDecls()) {
assert(VD->isCanonicalDecl() && "Expected canonical declaration");
Data.insert(std::make_pair(VD, MappedVarData()));
}
}
if (!NeedToDelayGlobalization) { if (!NeedToDelayGlobalization) {
emitGenericVarsProlog(CGF, D->getBeginLoc(), /*WithSPMDCheck=*/true); emitGenericVarsProlog(CGF, D->getBeginLoc(), /*WithSPMDCheck=*/true);
struct GlobalizationScope final : EHScopeStack::Cleanup { struct GlobalizationScope final : EHScopeStack::Cleanup {
GlobalizationScope() = default; GlobalizationScope() = default;
void Emit(CodeGenFunction &CGF, Flags flags) override { void Emit(CodeGenFunction &CGF, Flags flags) override {
static_cast<CGOpenMPRuntimeGPU &>(CGF.CGM.getOpenMPRuntime()) static_cast<CGOpenMPRuntimeGPU &>(CGF.CGM.getOpenMPRuntime())
.emitGenericVarsEpilog(CGF, /*WithSPMDCheck=*/true); .emitGenericVarsEpilog(CGF, /*WithSPMDCheck=*/true);
▲ Show 20 Lines • Show All 354 Lines • Show Last 20 Lines

clang/test/OpenMP/amdgcn_target_codegen.cpp

// REQUIRES: amdgpu-registered-target // REQUIRES: amdgpu-registered-target
// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -fopenmp-targets=amdgcn-amd-amdhsa -emit-llvm-bc %s -o %t-ppc-host.bc // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-unknown -fopenmp-targets=amdgcn-amd-amdhsa -emit-llvm-bc %s -o %t-ppc-host.bc
// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple amdgcn-amd-amdhsa -fopenmp-targets=amdgcn-amd-amdhsa -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple amdgcn-amd-amdhsa -fopenmp-targets=amdgcn-amd-amdhsa -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s
// expected-no-diagnostics // expected-no-diagnostics
#ifndef HEADER #ifndef HEADER
#define HEADER #define HEADER
#define N 1000 #define N 1000
int test_amdgcn_target_tid_threads() { int test_amdgcn_target_tid_threads() {
// CHECK-LABEL: define weak_odr protected amdgpu_kernel void @{{.*}}test_amdgcn_target_tid_threads // CHECK-LABEL: define weak_odr protected amdgpu_kernel void @{{.*}}test_amdgcn_target_tid_threads
int arr[N]; int arr[N];
// CHECK: call i32 @__kmpc_target_init(ptr addrspacecast (ptr addrspace(1) @1 to ptr), i8 1, i1 true, i1 true) // CHECK: call i32 @__kmpc_target_init(ptr addrspacecast (ptr addrspace(1) @1 to ptr), i8 1, i1 true)
#pragma omp target #pragma omp target
for (int i = 0; i < N; i++) { for (int i = 0; i < N; i++) {
arr[i] = 1; arr[i] = 1;
} }
return arr[0]; return arr[0];
} }
int test_amdgcn_target_tid_threads_simd() { int test_amdgcn_target_tid_threads_simd() {
// CHECK-LABEL: define weak_odr protected amdgpu_kernel void @{{.*}}test_amdgcn_target_tid_threads_simd // CHECK-LABEL: define weak_odr protected amdgpu_kernel void @{{.*}}test_amdgcn_target_tid_threads_simd
int arr[N]; int arr[N];
// CHECK: call i32 @__kmpc_target_init(ptr addrspacecast (ptr addrspace(1) @1 to ptr), i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init(ptr addrspacecast (ptr addrspace(1) @1 to ptr), i8 2, i1 false)
#pragma omp target simd #pragma omp target simd
for (int i = 0; i < N; i++) { for (int i = 0; i < N; i++) {
arr[i] = 1; arr[i] = 1;
} }
return arr[0]; return arr[0];
} }
#endif #endif

clang/test/OpenMP/declare_target_codegen_globalization.cpp

Show All 25 Lines
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z6maini1v_l16 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z6maini1v_l16
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP3]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP3]], align 8
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1)
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[B:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[B:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[CALL:%.*]] = call noundef i32 @_Z3fooRi(ptr noundef nonnull align 4 dereferenceable(4) [[B]]) #[[ATTR6:[0-9]+]] // CHECK1-NEXT: [[CALL:%.*]] = call noundef i32 @_Z3fooRi(ptr noundef nonnull align 4 dereferenceable(4) [[B]]) #[[ATTR7:[0-9]+]]
// CHECK1-NEXT: [[CALL1:%.*]] = call noundef i32 @_Z3barv() #[[ATTR6]] // CHECK1-NEXT: [[CALL1:%.*]] = call noundef i32 @_Z3barv() #[[ATTR7]]
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[CALL]], [[CALL1]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[CALL]], [[CALL1]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP0]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_Z3fooRi // CHECK1-LABEL: define {{[^@]+}}@_Z3fooRi
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: ret i32 [[TMP1]] // CHECK1-NEXT: ret i32 [[TMP1]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_Z3barv // CHECK1-LABEL: define {{[^@]+}}@_Z3barv
// CHECK1-SAME: () #[[ATTR2]] { // CHECK1-SAME: () #[[ATTR2]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 4) // CHECK1-NEXT: [[A:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 4)
// CHECK1-NEXT: [[CALL:%.*]] = call noundef i32 @_Z3fooRi(ptr noundef nonnull align 4 dereferenceable(4) [[A]]) #[[ATTR6]] // CHECK1-NEXT: [[CALL:%.*]] = call noundef i32 @_Z3fooRi(ptr noundef nonnull align 4 dereferenceable(4) [[A]]) #[[ATTR7]]
// CHECK1-NEXT: call void @__kmpc_free_shared(ptr [[A]], i64 4) // CHECK1-NEXT: call void @__kmpc_free_shared(ptr [[A]], i64 4)
// CHECK1-NEXT: ret i32 [[CALL]] // CHECK1-NEXT: ret i32 [[CALL]]
// //

clang/test/OpenMP/nvptx_SPMD_codegen.cpp

// Test target codegen - host bc file has to be created first. // Test target codegen - host bc file has to be created first.
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o %t-ppc-host.bc // RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm-bc %s -o %t-ppc-host.bc
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -x c++ -triple nvptx64-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s // RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -x c++ -triple nvptx64-unknown-unknown -fopenmp-targets=nvptx64-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-ppc-host.bc -o - | FileCheck %s
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=nvptx-nvidia-cuda -emit-llvm-bc %s -o %t-x86-host.bc // RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=nvptx-nvidia-cuda -emit-llvm-bc %s -o %t-x86-host.bc
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -x c++ -triple nvptx-unknown-unknown -fopenmp-targets=nvptx-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s // RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -x c++ -triple nvptx-unknown-unknown -fopenmp-targets=nvptx-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s
// RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -fexceptions -fcxx-exceptions -x c++ -triple nvptx-unknown-unknown -fopenmp-targets=nvptx-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s // RUN: %clang_cc1 -no-opaque-pointers -verify -fopenmp -fexceptions -fcxx-exceptions -x c++ -triple nvptx-unknown-unknown -fopenmp-targets=nvptx-nvidia-cuda -emit-llvm %s -fopenmp-is-device -fopenmp-host-ir-file-path %t-x86-host.bc -o - | FileCheck %s
// expected-no-diagnostics // expected-no-diagnostics
#ifndef HEADER #ifndef HEADER
#define HEADER #define HEADER
int a; int a;
// CHECK-NOT: @__omp_offloading_{{.+}}_exec_mode = weak constant i8 1 // CHECK-NOT: @__omp_offloading_{{.+}}_exec_mode = weak constant i8 1
// CHECK-DAG: [[DISTR_FULL:@.+]] = private unnamed_addr constant %struct.ident_t { i32 0, i32 2050, i32 1, i32 {{[0-9]+}}, i8* getelementptr inbounds // CHECK-DAG: [[DISTR_FULL:@.+]] = private unnamed_addr constant %struct.ident_t { i32 0, i32 2050, i32 0, i32 {{[0-9]+}}, i8* getelementptr inbounds
// CHECK-DAG: [[FULL:@.+]] = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 1, i32 {{[0-9]+}}, i8* getelementptr inbounds // CHECK-DAG: [[FULL:@.+]] = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 0, i32 {{[0-9]+}}, i8* getelementptr inbounds
// CHECK-DAG: [[BAR_FULL:@.+]] = private unnamed_addr constant %struct.ident_t { i32 0, i32 66, i32 1, i32 {{[0-9]+}}, i8* getelementptr inbounds // CHECK-DAG: [[BAR_FULL:@.+]] = private unnamed_addr constant %struct.ident_t { i32 0, i32 66, i32 0, i32 {{[0-9]+}}, i8* getelementptr inbounds
// CHECK-NOT: @__omp_offloading_{{.+}}_exec_mode = weak constant i8 1 // CHECK-NOT: @__omp_offloading_{{.+}}_exec_mode = weak constant i8 1
void foo() { void foo() {
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
#pragma omp target teams distribute parallel for simd if(a) #pragma omp target teams distribute parallel for simd if(a)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target teams distribute parallel for simd schedule(static) #pragma omp target teams distribute parallel for simd schedule(static)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target teams distribute parallel for simd schedule(static, 1) #pragma omp target teams distribute parallel for simd schedule(static, 1)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target teams distribute parallel for simd schedule(auto) #pragma omp target teams distribute parallel for simd schedule(auto)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target teams distribute parallel for simd schedule(runtime) #pragma omp target teams distribute parallel for simd schedule(runtime)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target teams distribute parallel for simd schedule(dynamic) #pragma omp target teams distribute parallel for simd schedule(dynamic)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target teams distribute parallel for simd schedule(guided) #pragma omp target teams distribute parallel for simd schedule(guided)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
int a; int a;
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
#pragma omp target teams distribute parallel for lastprivate(a) #pragma omp target teams distribute parallel for lastprivate(a)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
a = i; a = i;
#pragma omp target teams distribute parallel for schedule(static) #pragma omp target teams distribute parallel for schedule(static)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target teams distribute parallel for schedule(static, 1) #pragma omp target teams distribute parallel for schedule(static, 1)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target teams distribute parallel for schedule(auto) #pragma omp target teams distribute parallel for schedule(auto)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target teams distribute parallel for schedule(runtime) #pragma omp target teams distribute parallel for schedule(runtime)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target teams distribute parallel for schedule(dynamic) #pragma omp target teams distribute parallel for schedule(dynamic)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target teams distribute parallel for schedule(guided) #pragma omp target teams distribute parallel for schedule(guided)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init( // CHECK: call i32 @__kmpc_target_init(
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
#pragma omp target teams #pragma omp target teams
{ {
int b; int b;
#pragma omp distribute parallel for simd #pragma omp distribute parallel for simd
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
Show All 25 Lines
#pragma omp target teams #pragma omp target teams
#pragma omp distribute parallel for simd schedule(dynamic) #pragma omp distribute parallel for simd schedule(dynamic)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target teams #pragma omp target teams
#pragma omp distribute parallel for simd schedule(guided) #pragma omp distribute parallel for simd schedule(guided)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
#pragma omp target teams #pragma omp target teams
#pragma omp distribute parallel for #pragma omp distribute parallel for
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target teams #pragma omp target teams
#pragma omp distribute parallel for schedule(static) #pragma omp distribute parallel for schedule(static)
Show All 14 Lines
#pragma omp target teams #pragma omp target teams
#pragma omp distribute parallel for schedule(dynamic) #pragma omp distribute parallel for schedule(dynamic)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target teams #pragma omp target teams
#pragma omp distribute parallel for schedule(guided) #pragma omp distribute parallel for schedule(guided)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[DISTR_FULL]] // CHECK-DAG: [[DISTR_FULL]]
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
#pragma omp target #pragma omp target
#pragma omp teams #pragma omp teams
#pragma omp distribute parallel for #pragma omp distribute parallel for
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target #pragma omp target
#pragma omp teams #pragma omp teams
Show All 20 Lines
#pragma omp distribute parallel for schedule(dynamic) #pragma omp distribute parallel for schedule(dynamic)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target #pragma omp target
#pragma omp teams #pragma omp teams
#pragma omp distribute parallel for schedule(guided) #pragma omp distribute parallel for schedule(guided)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
#pragma omp target parallel for if(a) #pragma omp target parallel for if(a)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target parallel for schedule(static) #pragma omp target parallel for schedule(static)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target parallel for schedule(static, 1) #pragma omp target parallel for schedule(static, 1)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target parallel for schedule(auto) #pragma omp target parallel for schedule(auto)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target parallel for schedule(runtime) #pragma omp target parallel for schedule(runtime)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target parallel for schedule(dynamic) #pragma omp target parallel for schedule(dynamic)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target parallel for schedule(guided) #pragma omp target parallel for schedule(guided)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK-DAG: [[BAR_FULL]] // CHECK-DAG: [[BAR_FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK-DAG: [[BAR_FULL]] // CHECK-DAG: [[BAR_FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK-DAG: [[BAR_FULL]] // CHECK-DAG: [[BAR_FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK-DAG: [[BAR_FULL]] // CHECK-DAG: [[BAR_FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK-DAG: [[BAR_FULL]] // CHECK-DAG: [[BAR_FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK-DAG: [[BAR_FULL]] // CHECK-DAG: [[BAR_FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK-DAG: [[BAR_FULL]] // CHECK-DAG: [[BAR_FULL]]
#pragma omp target parallel if(a) #pragma omp target parallel if(a)
#pragma omp for simd #pragma omp for simd
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target parallel #pragma omp target parallel
#pragma omp for simd schedule(static) #pragma omp for simd schedule(static)
Show All 14 Lines
#pragma omp target parallel #pragma omp target parallel
#pragma omp for simd schedule(dynamic) #pragma omp for simd schedule(dynamic)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target parallel #pragma omp target parallel
#pragma omp for simd schedule(guided) #pragma omp for simd schedule(guided)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK-DAG: [[BAR_FULL]] // CHECK-DAG: [[BAR_FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK-DAG: [[BAR_FULL]] // CHECK-DAG: [[BAR_FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK-DAG: [[BAR_FULL]] // CHECK-DAG: [[BAR_FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK-DAG: [[BAR_FULL]] // CHECK-DAG: [[BAR_FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK-DAG: [[BAR_FULL]] // CHECK-DAG: [[BAR_FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK-DAG: [[BAR_FULL]] // CHECK-DAG: [[BAR_FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
#pragma omp target #pragma omp target
#pragma omp parallel #pragma omp parallel
#pragma omp for simd ordered #pragma omp for simd ordered
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target #pragma omp target
#pragma omp parallel #pragma omp parallel
Show All 20 Lines
#pragma omp for simd schedule(dynamic) #pragma omp for simd schedule(dynamic)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target #pragma omp target
#pragma omp parallel #pragma omp parallel
#pragma omp for simd schedule(guided) #pragma omp for simd schedule(guided)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-DAG: [[FULL]] // CHECK-DAG: [[FULL]]
#pragma omp target #pragma omp target
#pragma omp parallel for #pragma omp parallel for
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
; ;
#pragma omp target #pragma omp target
#pragma omp parallel for schedule(static) #pragma omp parallel for schedule(static)
for (int i = 0; i < 10; ++i) for (int i = 0; i < 10; ++i)
Show All 25 Lines

clang/test/OpenMP/nvptx_data_sharing.cpp

Show All 30 Lines
#endif #endif
// CHECK-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7test_dsv_l14 // CHECK-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7test_dsv_l14
// CHECK-SAME: () #[[ATTR0:[0-9]+]] { // CHECK-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK-NEXT: entry: // CHECK-NEXT: entry:
// CHECK-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8 // CHECK-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8
// CHECK-NEXT: [[C:%.*]] = alloca i32, align 4 // CHECK-NEXT: [[C:%.*]] = alloca i32, align 4
// CHECK-NEXT: [[CAPTURED_VARS_ADDRS1:%.*]] = alloca [2 x ptr], align 8 // CHECK-NEXT: [[CAPTURED_VARS_ADDRS1:%.*]] = alloca [2 x ptr], align 8
// CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // 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-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK: user_code.entry: // CHECK: user_code.entry:
// CHECK-NEXT: [[A:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 4) // CHECK-NEXT: [[A:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 4)
// CHECK-NEXT: [[B:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 4) // CHECK-NEXT: [[B:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 4)
// CHECK-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK-NEXT: store i32 10, ptr [[A]], align 4 // CHECK-NEXT: store i32 10, ptr [[A]], align 4
// CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK-NEXT: store ptr [[A]], ptr [[TMP2]], align 8 // CHECK-NEXT: store ptr [[A]], ptr [[TMP2]], align 8
// CHECK-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr @__omp_outlined___wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 1) // CHECK-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr @__omp_outlined___wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 1)
// CHECK-NEXT: store i32 100, ptr [[B]], align 4 // CHECK-NEXT: store i32 100, ptr [[B]], align 4
// CHECK-NEXT: store i32 1000, ptr [[C]], align 4 // CHECK-NEXT: store i32 1000, ptr [[C]], align 4
// CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS1]], i64 0, i64 0 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS1]], i64 0, i64 0
// CHECK-NEXT: store ptr [[B]], ptr [[TMP5]], align 8 // CHECK-NEXT: store ptr [[B]], ptr [[TMP3]], align 8
// CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS1]], i64 0, i64 1 // CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS1]], i64 0, i64 1
// CHECK-NEXT: store ptr [[A]], ptr [[TMP7]], align 8 // CHECK-NEXT: store ptr [[A]], ptr [[TMP4]], align 8
// CHECK-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS1]], i64 2) // CHECK-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS1]], i64 2)
// CHECK-NEXT: call void @__kmpc_free_shared(ptr [[B]], i64 4) // CHECK-NEXT: call void @__kmpc_free_shared(ptr [[B]], i64 4)
// CHECK-NEXT: call void @__kmpc_free_shared(ptr [[A]], i64 4) // CHECK-NEXT: call void @__kmpc_free_shared(ptr [[A]], i64 4)
// CHECK-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK-NEXT: ret void // CHECK-NEXT: ret void
// CHECK: worker.exit: // CHECK: worker.exit:
// CHECK-NEXT: ret void // CHECK-NEXT: ret void
// //
// //
// CHECK-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK-NEXT: entry: // CHECK-NEXT: entry:
Show All 16 Lines
// CHECK-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8 // CHECK-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8
// CHECK-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2 // CHECK-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2
// CHECK-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]]) // CHECK-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]])
// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8 // CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8
// CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0
// CHECK-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP3]], align 8 // CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 8
// CHECK-NEXT: call void @__omp_outlined__(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP5]]) #[[ATTR4:[0-9]+]] // CHECK-NEXT: call void @__omp_outlined__(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP4]]) #[[ATTR4:[0-9]+]]
// CHECK-NEXT: ret void // CHECK-NEXT: ret void
// //
// //
// CHECK-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { // CHECK-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK-NEXT: entry: // CHECK-NEXT: entry:
// CHECK-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
Show All 22 Lines
// CHECK-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8 // CHECK-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8
// CHECK-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2 // CHECK-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2
// CHECK-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]]) // CHECK-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]])
// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8 // CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8
// CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0
// CHECK-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP3]], align 8 // CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 8
// CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 1 // CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 1
// CHECK-NEXT: [[TMP8:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 8
// CHECK-NEXT: call void @__omp_outlined__1(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP5]], ptr [[TMP8]]) #[[ATTR4]] // CHECK-NEXT: call void @__omp_outlined__1(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP4]], ptr [[TMP6]]) #[[ATTR4]]
// CHECK-NEXT: ret void // CHECK-NEXT: ret void
// //

clang/test/OpenMP/nvptx_distribute_parallel_generic_mode_codegen.cpp

Show All 38 Lines
// CHECK4-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 // CHECK4-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8
// CHECK4-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK4-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK4-NEXT: store i64 [[ARGC]], ptr [[ARGC_ADDR]], align 8 // CHECK4-NEXT: store i64 [[ARGC]], ptr [[ARGC_ADDR]], align 8
// CHECK4-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 8 // CHECK4-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 8
// CHECK4-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8 // CHECK4-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8
// CHECK4-NEXT: [[TMP1:%.*]] = load ptr, ptr [[C_ADDR]], align 8 // CHECK4-NEXT: [[TMP1:%.*]] = load ptr, ptr [[C_ADDR]], align 8
// CHECK4-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK4-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK4-NEXT: [[TMP3:%.*]] = load ptr, ptr [[D_ADDR]], align 8 // CHECK4-NEXT: [[TMP3:%.*]] = load ptr, ptr [[D_ADDR]], align 8
// CHECK4-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK4-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK4-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP4]], -1 // CHECK4-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP4]], -1
// CHECK4-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK4-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK4: user_code.entry: // CHECK4: user_code.entry:
// CHECK4-NEXT: [[TMP5:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4:[0-9]+]]) // CHECK4-NEXT: [[TMP5:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK4-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK4-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK4-NEXT: store i32 [[TMP6]], ptr [[ARGC_CASTED]], align 4 // CHECK4-NEXT: store i32 [[TMP6]], ptr [[ARGC_CASTED]], align 4
// CHECK4-NEXT: [[TMP7:%.*]] = load i64, ptr [[ARGC_CASTED]], align 8 // CHECK4-NEXT: [[TMP7:%.*]] = load i64, ptr [[ARGC_CASTED]], align 8
// CHECK4-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK4-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK4-NEXT: store i32 [[TMP5]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK4-NEXT: store i32 [[TMP5]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK4-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]], ptr [[TMP1]], ptr [[TMP2]], i64 [[TMP7]], ptr [[TMP3]]) #[[ATTR5:[0-9]+]] // CHECK4-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]], ptr [[TMP1]], ptr [[TMP2]], i64 [[TMP7]], ptr [[TMP3]]) #[[ATTR5:[0-9]+]]
// CHECK4-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK4-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// CHECK4: worker.exit: // CHECK4: worker.exit:
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK4-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK4-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[C:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[D:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK4-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[C:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[D:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
▲ Show 20 Lines • Show All 42 Lines • ▼ Show 20 Lines
// CHECK4: omp.precond.then: // CHECK4: omp.precond.then:
// CHECK4-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK4-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK4-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK4-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK4-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK4-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK4-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 // CHECK4-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK4-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK4-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK4-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[B4]], ptr align 4 [[TMP0]], i64 40, i1 false) // CHECK4-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[B4]], ptr align 4 [[TMP0]], i64 40, i1 false)
// CHECK4-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() // CHECK4-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK4-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK4-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK4-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK4-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
// CHECK4-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP11]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]]) // CHECK4-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK4-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 // CHECK4-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK4-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK4-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK4-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] // CHECK4-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]]
// CHECK4-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK4-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK4: cond.true: // CHECK4: cond.true:
// CHECK4-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK4-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK4-NEXT: br label [[COND_END:%.*]] // CHECK4-NEXT: br label [[COND_END:%.*]]
// CHECK4: cond.false: // CHECK4: cond.false:
// CHECK4-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 // CHECK4-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK4-NEXT: br label [[COND_END]] // CHECK4-NEXT: br label [[COND_END]]
// CHECK4: cond.end: // CHECK4: cond.end:
// CHECK4-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] // CHECK4-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
// CHECK4-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK4-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK4-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK4-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK4-NEXT: store i32 [[TMP16]], ptr [[DOTOMP_IV]], align 4 // CHECK4-NEXT: store i32 [[TMP14]], ptr [[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: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK4-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK4-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK4-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], 1 // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], 1
// CHECK4-NEXT: [[CMP7:%.*]] = icmp slt i32 [[TMP17]], [[ADD]] // CHECK4-NEXT: [[CMP7:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
// CHECK4-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK4-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK4: omp.inner.for.body: // CHECK4: omp.inner.for.body:
// CHECK4-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK4-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK4-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64
// CHECK4-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK4-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64 // CHECK4-NEXT: [[TMP20:%.*]] = zext i32 [[TMP19]] to i64
// CHECK4-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 // CHECK4-NEXT: [[TMP21:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK4-NEXT: [[TMP22:%.*]] = zext i32 [[TMP21]] to i64 // CHECK4-NEXT: [[TMP22:%.*]] = inttoptr i64 [[TMP18]] to ptr
// CHECK4-NEXT: [[TMP23:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK4-NEXT: store ptr [[TMP22]], ptr [[TMP21]], align 8
// CHECK4-NEXT: [[TMP23:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK4-NEXT: [[TMP24:%.*]] = inttoptr i64 [[TMP20]] to ptr // CHECK4-NEXT: [[TMP24:%.*]] = inttoptr i64 [[TMP20]] to ptr
// CHECK4-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 8 // CHECK4-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 8
// CHECK4-NEXT: [[TMP25:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 // CHECK4-NEXT: [[TMP25:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK4-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP22]] to ptr // CHECK4-NEXT: store ptr [[ARGC_ADDR]], ptr [[TMP25]], align 8
// CHECK4-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8 // CHECK4-NEXT: [[TMP26:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2 // CHECK4-NEXT: store ptr [[TMP2]], ptr [[TMP26]], align 8
// CHECK4-NEXT: store ptr [[ARGC_ADDR]], ptr [[TMP27]], align 8 // CHECK4-NEXT: [[TMP27:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
// CHECK4-NEXT: [[TMP29:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3 // CHECK4-NEXT: store ptr [[B4]], ptr [[TMP27]], align 8
// CHECK4-NEXT: store ptr [[TMP2]], ptr [[TMP29]], align 8 // CHECK4-NEXT: [[TMP28:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 5
// CHECK4-NEXT: [[TMP31:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4 // CHECK4-NEXT: store ptr [[C1]], ptr [[TMP28]], align 8
// CHECK4-NEXT: store ptr [[B4]], ptr [[TMP31]], align 8 // CHECK4-NEXT: [[TMP29:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 6
// CHECK4-NEXT: [[TMP33:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 5 // CHECK4-NEXT: store ptr [[TMP3]], ptr [[TMP29]], align 8
// CHECK4-NEXT: store ptr [[C1]], ptr [[TMP33]], align 8 // CHECK4-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK4-NEXT: [[TMP35:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 6 // CHECK4-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP30]], 0
// CHECK4-NEXT: store ptr [[TMP3]], ptr [[TMP35]], align 8 // CHECK4-NEXT: [[TMP31:%.*]] = zext i1 [[TOBOOL]] to i32
// CHECK4-NEXT: [[TMP37:%.*]] = load i32, ptr [[TMP2]], align 4 // CHECK4-NEXT: [[TMP32:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK4-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP37]], 0 // CHECK4-NEXT: [[TMP33:%.*]] = load i32, ptr [[TMP32]], align 4
// CHECK4-NEXT: [[TMP38:%.*]] = zext i1 [[TOBOOL]] to i32 // CHECK4-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP33]], i32 [[TMP31]], i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 7)
// CHECK4-NEXT: [[TMP39:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK4-NEXT: [[TMP40:%.*]] = load i32, ptr [[TMP39]], align 4
// CHECK4-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP40]], i32 [[TMP38]], i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 7)
// 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: [[TMP42:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK4-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK4-NEXT: [[TMP43:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK4-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK4-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP42]], [[TMP43]] // CHECK4-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP34]], [[TMP35]]
// CHECK4-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4 // CHECK4-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4
// CHECK4-NEXT: [[TMP44:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK4-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK4-NEXT: [[TMP45:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK4-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK4-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP44]], [[TMP45]] // CHECK4-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP36]], [[TMP37]]
// CHECK4-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_COMB_LB]], align 4 // CHECK4-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_COMB_LB]], align 4
// CHECK4-NEXT: [[TMP46:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 // CHECK4-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK4-NEXT: [[TMP47:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK4-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK4-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP46]], [[TMP47]] // CHECK4-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP38]], [[TMP39]]
// CHECK4-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK4-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK4-NEXT: [[TMP48:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 // CHECK4-NEXT: [[TMP40:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK4-NEXT: [[TMP49:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK4-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK4-NEXT: [[CMP11:%.*]] = icmp sgt i32 [[TMP48]], [[TMP49]] // CHECK4-NEXT: [[CMP11:%.*]] = icmp sgt i32 [[TMP40]], [[TMP41]]
// CHECK4-NEXT: br i1 [[CMP11]], label [[COND_TRUE12:%.*]], label [[COND_FALSE13:%.*]] // CHECK4-NEXT: br i1 [[CMP11]], label [[COND_TRUE12:%.*]], label [[COND_FALSE13:%.*]]
// CHECK4: cond.true12: // CHECK4: cond.true12:
// CHECK4-NEXT: [[TMP50:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK4-NEXT: [[TMP42:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK4-NEXT: br label [[COND_END14:%.*]] // CHECK4-NEXT: br label [[COND_END14:%.*]]
// CHECK4: cond.false13: // CHECK4: cond.false13:
// CHECK4-NEXT: [[TMP51:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 // CHECK4-NEXT: [[TMP43:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK4-NEXT: br label [[COND_END14]] // CHECK4-NEXT: br label [[COND_END14]]
// CHECK4: cond.end14: // CHECK4: cond.end14:
// CHECK4-NEXT: [[COND15:%.*]] = phi i32 [ [[TMP50]], [[COND_TRUE12]] ], [ [[TMP51]], [[COND_FALSE13]] ] // CHECK4-NEXT: [[COND15:%.*]] = phi i32 [ [[TMP42]], [[COND_TRUE12]] ], [ [[TMP43]], [[COND_FALSE13]] ]
// CHECK4-NEXT: store i32 [[COND15]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK4-NEXT: store i32 [[COND15]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK4-NEXT: [[TMP52:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK4-NEXT: [[TMP44:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK4-NEXT: store i32 [[TMP52]], ptr [[DOTOMP_IV]], align 4 // CHECK4-NEXT: store i32 [[TMP44]], ptr [[DOTOMP_IV]], align 4
// CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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: [[TMP53:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK4-NEXT: [[TMP45:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK4-NEXT: [[TMP54:%.*]] = load i32, ptr [[TMP53]], align 4 // CHECK4-NEXT: [[TMP46:%.*]] = load i32, ptr [[TMP45]], align 4
// CHECK4-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP54]]) // CHECK4-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP46]])
// CHECK4-NEXT: [[TMP55:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK4-NEXT: [[TMP47:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK4-NEXT: [[TMP56:%.*]] = icmp ne i32 [[TMP55]], 0 // CHECK4-NEXT: [[TMP48:%.*]] = icmp ne i32 [[TMP47]], 0
// CHECK4-NEXT: br i1 [[TMP56]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]] // CHECK4-NEXT: br i1 [[TMP48]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
// CHECK4: .omp.lastprivate.then: // CHECK4: .omp.lastprivate.then:
// CHECK4-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP1]], ptr align 4 [[C1]], i64 40, i1 false) // CHECK4-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP1]], ptr align 4 [[C1]], i64 40, i1 false)
// CHECK4-NEXT: br label [[DOTOMP_LASTPRIVATE_DONE]] // CHECK4-NEXT: br label [[DOTOMP_LASTPRIVATE_DONE]]
// CHECK4: .omp.lastprivate.done: // CHECK4: .omp.lastprivate.done:
// CHECK4-NEXT: br label [[OMP_PRECOND_END]] // CHECK4-NEXT: br label [[OMP_PRECOND_END]]
// CHECK4: omp.precond.end: // CHECK4: omp.precond.end:
// CHECK4-NEXT: call void @__kmpc_free_shared(ptr [[C1]], i64 40) // CHECK4-NEXT: call void @__kmpc_free_shared(ptr [[C1]], i64 40)
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
▲ Show 20 Lines • Show All 56 Lines • ▼ Show 20 Lines
// CHECK4-NEXT: [[CONV:%.*]] = trunc i64 [[TMP9]] to i32 // CHECK4-NEXT: [[CONV:%.*]] = trunc i64 [[TMP9]] to i32
// CHECK4-NEXT: [[TMP10:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK4-NEXT: [[TMP10:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK4-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP10]] to i32 // CHECK4-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP10]] to i32
// CHECK4-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4 // CHECK4-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4
// CHECK4-NEXT: store i32 [[CONV3]], ptr [[DOTOMP_UB]], align 4 // CHECK4-NEXT: store i32 [[CONV3]], ptr [[DOTOMP_UB]], align 4
// CHECK4-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 // CHECK4-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK4-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK4-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK4-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[B4]], ptr align 4 [[TMP2]], i64 40, i1 false) // CHECK4-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[B4]], ptr align 4 [[TMP2]], i64 40, i1 false)
// CHECK4-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK4-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK4-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP13]], align 4 // CHECK4-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
// CHECK4-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP14]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK4-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP12]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK4-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK4-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK4-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4 // CHECK4-NEXT: store i32 [[TMP13]], ptr [[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: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK4-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK4-NEXT: [[CONV7:%.*]] = sext i32 [[TMP16]] to i64 // CHECK4-NEXT: [[CONV7:%.*]] = sext i32 [[TMP14]] to i64
// CHECK4-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK4-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK4-NEXT: [[CMP8:%.*]] = icmp ule i64 [[CONV7]], [[TMP17]] // CHECK4-NEXT: [[CMP8:%.*]] = icmp ule i64 [[CONV7]], [[TMP15]]
// CHECK4-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK4-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK4: omp.inner.for.body: // CHECK4: omp.inner.for.body:
// CHECK4-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK4-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK4-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1 // CHECK4-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1
// CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK4-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK4-NEXT: store i32 [[ADD]], ptr [[I6]], align 4 // CHECK4-NEXT: store i32 [[ADD]], ptr [[I6]], align 4
// CHECK4-NEXT: [[CALL:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[I6]]) #[[ATTR8:[0-9]+]] // CHECK4-NEXT: [[CALL:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[I6]]) #[[ATTR8:[0-9]+]]
// CHECK4-NEXT: [[CALL9:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[TMP1]]) #[[ATTR8]] // CHECK4-NEXT: [[CALL9:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[TMP1]]) #[[ATTR8]]
// CHECK4-NEXT: [[ADD10:%.*]] = add nsw i32 [[CALL]], [[CALL9]] // CHECK4-NEXT: [[ADD10:%.*]] = add nsw i32 [[CALL]], [[CALL9]]
// CHECK4-NEXT: [[TMP19:%.*]] = load i32, ptr [[I6]], align 4 // CHECK4-NEXT: [[TMP17:%.*]] = load i32, ptr [[I6]], align 4
// CHECK4-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP19]] to i64 // CHECK4-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP17]] to i64
// CHECK4-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[B4]], i64 0, i64 [[IDXPROM]] // CHECK4-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[B4]], i64 0, i64 [[IDXPROM]]
// CHECK4-NEXT: [[CALL11:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[ARRAYIDX]]) #[[ATTR8]] // CHECK4-NEXT: [[CALL11:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[ARRAYIDX]]) #[[ATTR8]]
// CHECK4-NEXT: [[ADD12:%.*]] = add nsw i32 [[ADD10]], [[CALL11]] // CHECK4-NEXT: [[ADD12:%.*]] = add nsw i32 [[ADD10]], [[CALL11]]
// CHECK4-NEXT: [[TMP20:%.*]] = load i32, ptr [[I6]], align 4 // CHECK4-NEXT: [[TMP18:%.*]] = load i32, ptr [[I6]], align 4
// CHECK4-NEXT: [[IDXPROM13:%.*]] = sext i32 [[TMP20]] to i64 // CHECK4-NEXT: [[IDXPROM13:%.*]] = sext i32 [[TMP18]] to i64
// CHECK4-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds [10 x i32], ptr [[C5]], i64 0, i64 [[IDXPROM13]] // CHECK4-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds [10 x i32], ptr [[C5]], i64 0, i64 [[IDXPROM13]]
// CHECK4-NEXT: [[CALL15:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[ARRAYIDX14]]) #[[ATTR8]] // CHECK4-NEXT: [[CALL15:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[ARRAYIDX14]]) #[[ATTR8]]
// CHECK4-NEXT: [[ADD16:%.*]] = add nsw i32 [[ADD12]], [[CALL15]] // CHECK4-NEXT: [[ADD16:%.*]] = add nsw i32 [[ADD12]], [[CALL15]]
// CHECK4-NEXT: [[TMP21:%.*]] = load i32, ptr [[I6]], align 4 // CHECK4-NEXT: [[TMP19:%.*]] = load i32, ptr [[I6]], align 4
// CHECK4-NEXT: [[IDXPROM17:%.*]] = sext i32 [[TMP21]] to i64 // CHECK4-NEXT: [[IDXPROM17:%.*]] = sext i32 [[TMP19]] to i64
// CHECK4-NEXT: [[ARRAYIDX18:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP4]], i64 0, i64 [[IDXPROM17]] // CHECK4-NEXT: [[ARRAYIDX18:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP4]], i64 0, i64 [[IDXPROM17]]
// CHECK4-NEXT: [[CALL19:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[ARRAYIDX18]]) #[[ATTR8]] // CHECK4-NEXT: [[CALL19:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[ARRAYIDX18]]) #[[ATTR8]]
// CHECK4-NEXT: [[ADD20:%.*]] = add nsw i32 [[ADD16]], [[CALL19]] // CHECK4-NEXT: [[ADD20:%.*]] = add nsw i32 [[ADD16]], [[CALL19]]
// CHECK4-NEXT: store i32 [[ADD20]], ptr [[TMP1]], align 4 // CHECK4-NEXT: store i32 [[ADD20]], ptr [[TMP1]], align 4
// 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: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK4-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK4-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK4-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK4-NEXT: [[ADD21:%.*]] = add nsw i32 [[TMP22]], [[TMP23]] // CHECK4-NEXT: [[ADD21:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
// CHECK4-NEXT: store i32 [[ADD21]], ptr [[DOTOMP_IV]], align 4 // CHECK4-NEXT: store i32 [[ADD21]], ptr [[DOTOMP_IV]], align 4
// CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK4-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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: [[TMP24:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK4-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK4-NEXT: [[TMP25:%.*]] = load i32, ptr [[TMP24]], align 4 // CHECK4-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
// CHECK4-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP25]]) // CHECK4-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP23]])
// CHECK4-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK4-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK4-NEXT: [[TMP27:%.*]] = icmp ne i32 [[TMP26]], 0 // CHECK4-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
// CHECK4-NEXT: br i1 [[TMP27]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]] // CHECK4-NEXT: br i1 [[TMP25]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
// CHECK4: .omp.lastprivate.then: // CHECK4: .omp.lastprivate.then:
// CHECK4-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP3]], ptr align 4 [[C5]], i64 40, i1 false) // CHECK4-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP3]], ptr align 4 [[C5]], i64 40, i1 false)
// CHECK4-NEXT: br label [[DOTOMP_LASTPRIVATE_DONE]] // CHECK4-NEXT: br label [[DOTOMP_LASTPRIVATE_DONE]]
// CHECK4: .omp.lastprivate.done: // CHECK4: .omp.lastprivate.done:
// CHECK4-NEXT: br label [[OMP_PRECOND_END]] // CHECK4-NEXT: br label [[OMP_PRECOND_END]]
// CHECK4: omp.precond.end: // CHECK4: omp.precond.end:
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
Show All 13 Lines
// CHECK5-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 4 // CHECK5-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 4
// CHECK5-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 // CHECK5-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK5-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK5-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK5-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 4 // CHECK5-NEXT: store ptr [[D]], ptr [[D_ADDR]], align 4
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 4
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[C_ADDR]], align 4 // CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[C_ADDR]], align 4
// CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4 // CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[D_ADDR]], align 4 // CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[D_ADDR]], align 4
// CHECK5-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK5-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK5-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP4]], -1 // CHECK5-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP4]], -1
// CHECK5-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK5-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK5: user_code.entry: // CHECK5: user_code.entry:
// CHECK5-NEXT: [[TMP5:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4:[0-9]+]]) // CHECK5-NEXT: [[TMP5:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK5-NEXT: store i32 [[TMP6]], ptr [[ARGC_CASTED]], align 4 // CHECK5-NEXT: store i32 [[TMP6]], ptr [[ARGC_CASTED]], align 4
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARGC_CASTED]], align 4 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARGC_CASTED]], align 4
// CHECK5-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK5-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK5-NEXT: store i32 [[TMP5]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK5-NEXT: store i32 [[TMP5]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK5-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]], ptr [[TMP1]], ptr [[TMP2]], i32 [[TMP7]], ptr [[TMP3]]) #[[ATTR5:[0-9]+]] // CHECK5-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]], ptr [[TMP1]], ptr [[TMP2]], i32 [[TMP7]], ptr [[TMP3]]) #[[ATTR5:[0-9]+]]
// CHECK5-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK5-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// CHECK5: worker.exit: // CHECK5: worker.exit:
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK5-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[C:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[D:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[C:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[D:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
▲ Show 20 Lines • Show All 42 Lines • ▼ Show 20 Lines
// CHECK5: omp.precond.then: // CHECK5: omp.precond.then:
// CHECK5-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK5-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[B4]], ptr align 4 [[TMP0]], i32 40, i1 false) // CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[B4]], ptr align 4 [[TMP0]], i32 40, i1 false)
// CHECK5-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() // CHECK5-NEXT: [[NVPTX_NUM_THREADS:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
// CHECK5-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
// CHECK5-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP11]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]]) // CHECK5-NEXT: call void @__kmpc_distribute_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[NVPTX_NUM_THREADS]])
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK5-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP12]], [[TMP13]] // CHECK5-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]]
// CHECK5-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK5-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK5: cond.true: // CHECK5: cond.true:
// CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK5-NEXT: br label [[COND_END:%.*]] // CHECK5-NEXT: br label [[COND_END:%.*]]
// CHECK5: cond.false: // CHECK5: cond.false:
// CHECK5-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: br label [[COND_END]] // CHECK5-NEXT: br label [[COND_END]]
// CHECK5: cond.end: // CHECK5: cond.end:
// CHECK5-NEXT: [[COND:%.*]] = phi i32 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
// CHECK5-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP16]], ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond: // CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK5-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], 1 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], 1
// CHECK5-NEXT: [[CMP7:%.*]] = icmp slt i32 [[TMP17]], [[ADD]] // CHECK5-NEXT: [[CMP7:%.*]] = icmp slt i32 [[TMP15]], [[ADD]]
// CHECK5-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body: // CHECK5: omp.inner.for.body:
// CHECK5-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP21:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0 // CHECK5-NEXT: [[TMP19:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK5-NEXT: [[TMP22:%.*]] = inttoptr i32 [[TMP19]] to ptr // CHECK5-NEXT: [[TMP20:%.*]] = inttoptr i32 [[TMP17]] to ptr
// CHECK5-NEXT: store ptr [[TMP20]], ptr [[TMP19]], align 4
// CHECK5-NEXT: [[TMP21:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK5-NEXT: [[TMP22:%.*]] = inttoptr i32 [[TMP18]] to ptr
// CHECK5-NEXT: store ptr [[TMP22]], ptr [[TMP21]], align 4 // CHECK5-NEXT: store ptr [[TMP22]], ptr [[TMP21]], align 4
// CHECK5-NEXT: [[TMP23:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1 // CHECK5-NEXT: [[TMP23:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
// CHECK5-NEXT: [[TMP24:%.*]] = inttoptr i32 [[TMP20]] to ptr // CHECK5-NEXT: store ptr [[ARGC_ADDR]], ptr [[TMP23]], align 4
// CHECK5-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 4 // CHECK5-NEXT: [[TMP24:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
// CHECK5-NEXT: [[TMP25:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2 // CHECK5-NEXT: store ptr [[TMP2]], ptr [[TMP24]], align 4
// CHECK5-NEXT: store ptr [[ARGC_ADDR]], ptr [[TMP25]], align 4 // CHECK5-NEXT: [[TMP25:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 4
// CHECK5-NEXT: [[TMP27:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3 // CHECK5-NEXT: store ptr [[B4]], ptr [[TMP25]], align 4
// CHECK5-NEXT: store ptr [[TMP2]], ptr [[TMP27]], align 4 // CHECK5-NEXT: [[TMP26:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 5
// CHECK5-NEXT: [[TMP29:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 4 // CHECK5-NEXT: store ptr [[C1]], ptr [[TMP26]], align 4
// CHECK5-NEXT: store ptr [[B4]], ptr [[TMP29]], align 4 // CHECK5-NEXT: [[TMP27:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 6
// CHECK5-NEXT: [[TMP31:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 5 // CHECK5-NEXT: store ptr [[TMP3]], ptr [[TMP27]], align 4
// CHECK5-NEXT: store ptr [[C1]], ptr [[TMP31]], align 4 // CHECK5-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK5-NEXT: [[TMP33:%.*]] = getelementptr inbounds [7 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 6 // CHECK5-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP28]], 0
// CHECK5-NEXT: store ptr [[TMP3]], ptr [[TMP33]], align 4 // CHECK5-NEXT: [[TMP29:%.*]] = zext i1 [[TOBOOL]] to i32
// CHECK5-NEXT: [[TMP35:%.*]] = load i32, ptr [[TMP2]], align 4 // CHECK5-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK5-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP35]], 0 // CHECK5-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
// CHECK5-NEXT: [[TMP36:%.*]] = zext i1 [[TOBOOL]] to i32 // CHECK5-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP31]], i32 [[TMP29]], i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 7)
// CHECK5-NEXT: [[TMP37:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK5-NEXT: [[TMP38:%.*]] = load i32, ptr [[TMP37]], align 4
// CHECK5-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP38]], i32 [[TMP36]], i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 7)
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc: // CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP40:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP40]], [[TMP41]] // CHECK5-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP32]], [[TMP33]]
// CHECK5-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP42:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: [[TMP43:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP42]], [[TMP43]] // CHECK5-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP34]], [[TMP35]]
// CHECK5-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: [[TMP44:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP45:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: [[TMP37:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP44]], [[TMP45]] // CHECK5-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP36]], [[TMP37]]
// CHECK5-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP46:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP38:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP47:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK5-NEXT: [[TMP39:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK5-NEXT: [[CMP11:%.*]] = icmp sgt i32 [[TMP46]], [[TMP47]] // CHECK5-NEXT: [[CMP11:%.*]] = icmp sgt i32 [[TMP38]], [[TMP39]]
// CHECK5-NEXT: br i1 [[CMP11]], label [[COND_TRUE12:%.*]], label [[COND_FALSE13:%.*]] // CHECK5-NEXT: br i1 [[CMP11]], label [[COND_TRUE12:%.*]], label [[COND_FALSE13:%.*]]
// CHECK5: cond.true12: // CHECK5: cond.true12:
// CHECK5-NEXT: [[TMP48:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK5-NEXT: [[TMP40:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK5-NEXT: br label [[COND_END14:%.*]] // CHECK5-NEXT: br label [[COND_END14:%.*]]
// CHECK5: cond.false13: // CHECK5: cond.false13:
// CHECK5-NEXT: [[TMP49:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: br label [[COND_END14]] // CHECK5-NEXT: br label [[COND_END14]]
// CHECK5: cond.end14: // CHECK5: cond.end14:
// CHECK5-NEXT: [[COND15:%.*]] = phi i32 [ [[TMP48]], [[COND_TRUE12]] ], [ [[TMP49]], [[COND_FALSE13]] ] // CHECK5-NEXT: [[COND15:%.*]] = phi i32 [ [[TMP40]], [[COND_TRUE12]] ], [ [[TMP41]], [[COND_FALSE13]] ]
// CHECK5-NEXT: store i32 [[COND15]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: store i32 [[COND15]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP50:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: [[TMP42:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP50]], ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: store i32 [[TMP42]], ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK5: omp.inner.for.end: // CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit: // CHECK5: omp.loop.exit:
// CHECK5-NEXT: [[TMP51:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK5-NEXT: [[TMP43:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK5-NEXT: [[TMP52:%.*]] = load i32, ptr [[TMP51]], align 4 // CHECK5-NEXT: [[TMP44:%.*]] = load i32, ptr [[TMP43]], align 4
// CHECK5-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP52]]) // CHECK5-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP44]])
// CHECK5-NEXT: [[TMP53:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP45:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP54:%.*]] = icmp ne i32 [[TMP53]], 0 // CHECK5-NEXT: [[TMP46:%.*]] = icmp ne i32 [[TMP45]], 0
// CHECK5-NEXT: br i1 [[TMP54]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]] // CHECK5-NEXT: br i1 [[TMP46]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
// CHECK5: .omp.lastprivate.then: // CHECK5: .omp.lastprivate.then:
// CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP1]], ptr align 4 [[C1]], i32 40, i1 false) // CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP1]], ptr align 4 [[C1]], i32 40, i1 false)
// CHECK5-NEXT: br label [[DOTOMP_LASTPRIVATE_DONE]] // CHECK5-NEXT: br label [[DOTOMP_LASTPRIVATE_DONE]]
// CHECK5: .omp.lastprivate.done: // CHECK5: .omp.lastprivate.done:
// CHECK5-NEXT: br label [[OMP_PRECOND_END]] // CHECK5-NEXT: br label [[OMP_PRECOND_END]]
// CHECK5: omp.precond.end: // CHECK5: omp.precond.end:
// CHECK5-NEXT: call void @__kmpc_free_shared(ptr [[C1]], i32 40) // CHECK5-NEXT: call void @__kmpc_free_shared(ptr [[C1]], i32 40)
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_UB]], align 4 // CHECK5-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4
// CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK5-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 [[TMP9]], ptr [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP10]], ptr [[DOTOMP_UB]], align 4 // CHECK5-NEXT: store i32 [[TMP10]], ptr [[DOTOMP_UB]], align 4
// CHECK5-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[B3]], ptr align 4 [[TMP2]], i32 40, i1 false) // CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[B3]], ptr align 4 [[TMP2]], i32 40, i1 false)
// CHECK5-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK5-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP13]], align 4 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
// CHECK5-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP14]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) // CHECK5-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP12]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK5-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond: // CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4 // CHECK5-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4
// CHECK5-NEXT: [[CMP6:%.*]] = icmp ule i32 [[TMP16]], [[TMP17]] // CHECK5-NEXT: [[CMP6:%.*]] = icmp ule i32 [[TMP14]], [[TMP15]]
// CHECK5-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body: // CHECK5: omp.inner.for.body:
// CHECK5-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[I5]], align 4 // CHECK5-NEXT: store i32 [[ADD]], ptr [[I5]], align 4
// CHECK5-NEXT: [[CALL:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[I5]]) #[[ATTR8:[0-9]+]] // CHECK5-NEXT: [[CALL:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[I5]]) #[[ATTR8:[0-9]+]]
// CHECK5-NEXT: [[CALL7:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[TMP1]]) #[[ATTR8]] // CHECK5-NEXT: [[CALL7:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[TMP1]]) #[[ATTR8]]
// CHECK5-NEXT: [[ADD8:%.*]] = add nsw i32 [[CALL]], [[CALL7]] // CHECK5-NEXT: [[ADD8:%.*]] = add nsw i32 [[CALL]], [[CALL7]]
// CHECK5-NEXT: [[TMP19:%.*]] = load i32, ptr [[I5]], align 4 // CHECK5-NEXT: [[TMP17:%.*]] = load i32, ptr [[I5]], align 4
// CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[B3]], i32 0, i32 [[TMP19]] // CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[B3]], i32 0, i32 [[TMP17]]
// CHECK5-NEXT: [[CALL9:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[ARRAYIDX]]) #[[ATTR8]] // CHECK5-NEXT: [[CALL9:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[ARRAYIDX]]) #[[ATTR8]]
// CHECK5-NEXT: [[ADD10:%.*]] = add nsw i32 [[ADD8]], [[CALL9]] // CHECK5-NEXT: [[ADD10:%.*]] = add nsw i32 [[ADD8]], [[CALL9]]
// CHECK5-NEXT: [[TMP20:%.*]] = load i32, ptr [[I5]], align 4 // CHECK5-NEXT: [[TMP18:%.*]] = load i32, ptr [[I5]], align 4
// CHECK5-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [10 x i32], ptr [[C4]], i32 0, i32 [[TMP20]] // CHECK5-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [10 x i32], ptr [[C4]], i32 0, i32 [[TMP18]]
// CHECK5-NEXT: [[CALL12:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[ARRAYIDX11]]) #[[ATTR8]] // CHECK5-NEXT: [[CALL12:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[ARRAYIDX11]]) #[[ATTR8]]
// CHECK5-NEXT: [[ADD13:%.*]] = add nsw i32 [[ADD10]], [[CALL12]] // CHECK5-NEXT: [[ADD13:%.*]] = add nsw i32 [[ADD10]], [[CALL12]]
// CHECK5-NEXT: [[TMP21:%.*]] = load i32, ptr [[I5]], align 4 // CHECK5-NEXT: [[TMP19:%.*]] = load i32, ptr [[I5]], align 4
// CHECK5-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP4]], i32 0, i32 [[TMP21]] // CHECK5-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP4]], i32 0, i32 [[TMP19]]
// CHECK5-NEXT: [[CALL15:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[ARRAYIDX14]]) #[[ATTR8]] // CHECK5-NEXT: [[CALL15:%.*]] = call noundef i32 @_Z3fooPi(ptr noundef [[ARRAYIDX14]]) #[[ATTR8]]
// CHECK5-NEXT: [[ADD16:%.*]] = add nsw i32 [[ADD13]], [[CALL15]] // CHECK5-NEXT: [[ADD16:%.*]] = add nsw i32 [[ADD13]], [[CALL15]]
// CHECK5-NEXT: store i32 [[ADD16]], ptr [[TMP1]], align 4 // CHECK5-NEXT: store i32 [[ADD16]], ptr [[TMP1]], align 4
// CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK5: omp.body.continue: // CHECK5: omp.body.continue:
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc: // CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK5-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK5-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP22]], [[TMP23]] // CHECK5-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
// CHECK5-NEXT: store i32 [[ADD17]], ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: store i32 [[ADD17]], ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK5: omp.inner.for.end: // CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit: // CHECK5: omp.loop.exit:
// CHECK5-NEXT: [[TMP24:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK5-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK5-NEXT: [[TMP25:%.*]] = load i32, ptr [[TMP24]], align 4 // CHECK5-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
// CHECK5-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP25]]) // CHECK5-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP23]])
// CHECK5-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP27:%.*]] = icmp ne i32 [[TMP26]], 0 // CHECK5-NEXT: [[TMP25:%.*]] = icmp ne i32 [[TMP24]], 0
// CHECK5-NEXT: br i1 [[TMP27]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]] // CHECK5-NEXT: br i1 [[TMP25]], label [[DOTOMP_LASTPRIVATE_THEN:%.*]], label [[DOTOMP_LASTPRIVATE_DONE:%.*]]
// CHECK5: .omp.lastprivate.then: // CHECK5: .omp.lastprivate.then:
// CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP3]], ptr align 4 [[C4]], i32 40, i1 false) // CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[TMP3]], ptr align 4 [[C4]], i32 40, i1 false)
// CHECK5-NEXT: br label [[DOTOMP_LASTPRIVATE_DONE]] // CHECK5-NEXT: br label [[DOTOMP_LASTPRIVATE_DONE]]
// CHECK5: .omp.lastprivate.done: // CHECK5: .omp.lastprivate.done:
// CHECK5-NEXT: br label [[OMP_PRECOND_END]] // CHECK5-NEXT: br label [[OMP_PRECOND_END]]
// CHECK5: omp.precond.end: // CHECK5: omp.precond.end:
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //

clang/test/OpenMP/nvptx_lambda_capturing.cpp

Show First 20 Lines • Show All 116 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP15]], i32 0, i32 3 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP15]], i32 0, i32 3
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP15]], i32 0, i32 3 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP15]], i32 0, i32 3
// CHECK1-NEXT: [[TMP27:%.*]] = load ptr, ptr [[TMP26]], align 8 // CHECK1-NEXT: [[TMP27:%.*]] = load ptr, ptr [[TMP26]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP15]], i32 0, i32 4 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP15]], i32 0, i32 4
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP15]], i32 0, i32 4 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP15]], i32 0, i32 4
// CHECK1-NEXT: [[TMP30:%.*]] = load ptr, ptr [[TMP29]], align 8 // CHECK1-NEXT: [[TMP30:%.*]] = load ptr, ptr [[TMP29]], align 8
// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store i64 [[TMP11]], ptr [[TMP31]], align 8 // CHECK1-NEXT: store i64 [[TMP11]], ptr [[TMP31]], align 8
// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store i64 [[TMP11]], ptr [[TMP33]], align 8 // CHECK1-NEXT: store i64 [[TMP11]], ptr [[TMP32]], align 8
// CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP35]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP33]], align 8
// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP36]], align 8 // CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP34]], align 8
// CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP38]], align 8 // CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP35]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store ptr null, ptr [[TMP40]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP36]], align 8
// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP41]], align 8 // CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP37]], align 8
// CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP43]], align 8 // CHECK1-NEXT: store ptr [[TMP13]], ptr [[TMP38]], align 8
// CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: store ptr null, ptr [[TMP39]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP40]], align 8
// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP41]], align 8
// CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK1-NEXT: store ptr null, ptr [[TMP42]], align 8
// CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK1-NEXT: store ptr [[A]], ptr [[TMP43]], align 8
// CHECK1-NEXT: [[TMP44:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK1-NEXT: store ptr [[A]], ptr [[TMP44]], align 8
// CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK1-NEXT: store ptr null, ptr [[TMP45]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP45]], align 8
// CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5
// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP46]], align 8 // CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP46]], align 8
// CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 5
// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP48]], align 8 // CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP47]], align 8
// CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5
// CHECK1-NEXT: store ptr null, ptr [[TMP50]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP48]], align 8
// CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6
// CHECK1-NEXT: store ptr [[A]], ptr [[TMP51]], align 8 // CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP49]], align 8
// CHECK1-NEXT: [[TMP53:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 6
// CHECK1-NEXT: store ptr [[A]], ptr [[TMP53]], align 8 // CHECK1-NEXT: store ptr [[TMP18]], ptr [[TMP50]], align 8
// CHECK1-NEXT: [[TMP55:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 // CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 6
// CHECK1-NEXT: store ptr null, ptr [[TMP55]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP51]], align 8
// CHECK1-NEXT: [[TMP56:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP52:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7
// CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP56]], align 8 // CHECK1-NEXT: store ptr [[TMP19]], ptr [[TMP52]], align 8
// CHECK1-NEXT: [[TMP58:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP53:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 7
// CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP58]], align 8 // CHECK1-NEXT: store ptr [[TMP21]], ptr [[TMP53]], align 8
// CHECK1-NEXT: [[TMP60:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 5 // CHECK1-NEXT: [[TMP54:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 7
// CHECK1-NEXT: store ptr null, ptr [[TMP54]], align 8
// CHECK1-NEXT: [[TMP55:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 8
// CHECK1-NEXT: store ptr [[TMP22]], ptr [[TMP55]], align 8
// CHECK1-NEXT: [[TMP56:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 8
// CHECK1-NEXT: store ptr [[TMP24]], ptr [[TMP56]], align 8
// CHECK1-NEXT: [[TMP57:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 8
// CHECK1-NEXT: store ptr null, ptr [[TMP57]], align 8
// CHECK1-NEXT: [[TMP58:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 9
// CHECK1-NEXT: store ptr [[TMP25]], ptr [[TMP58]], align 8
// CHECK1-NEXT: [[TMP59:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 9
// CHECK1-NEXT: store ptr [[TMP27]], ptr [[TMP59]], align 8
// CHECK1-NEXT: [[TMP60:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 9
// CHECK1-NEXT: store ptr null, ptr [[TMP60]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP60]], align 8
// CHECK1-NEXT: [[TMP61:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP61:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 10
// CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP61]], align 8 // CHECK1-NEXT: store ptr [[TMP28]], ptr [[TMP61]], align 8
// CHECK1-NEXT: [[TMP63:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP62:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 10
// CHECK1-NEXT: store ptr [[TMP18]], ptr [[TMP63]], align 8 // CHECK1-NEXT: store ptr [[TMP30]], ptr [[TMP62]], align 8
// CHECK1-NEXT: [[TMP65:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 6 // CHECK1-NEXT: [[TMP63:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 10
// CHECK1-NEXT: store ptr null, ptr [[TMP65]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP63]], align 8
// CHECK1-NEXT: [[TMP66:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP64:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[TMP19]], ptr [[TMP66]], align 8 // CHECK1-NEXT: [[TMP65:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP68:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 7
// CHECK1-NEXT: store ptr [[TMP21]], ptr [[TMP68]], align 8
// CHECK1-NEXT: [[TMP70:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 7
// CHECK1-NEXT: store ptr null, ptr [[TMP70]], align 8
// CHECK1-NEXT: [[TMP71:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 8
// CHECK1-NEXT: store ptr [[TMP22]], ptr [[TMP71]], align 8
// CHECK1-NEXT: [[TMP73:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 8
// CHECK1-NEXT: store ptr [[TMP24]], ptr [[TMP73]], align 8
// CHECK1-NEXT: [[TMP75:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 8
// CHECK1-NEXT: store ptr null, ptr [[TMP75]], align 8
// CHECK1-NEXT: [[TMP76:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 9
// CHECK1-NEXT: store ptr [[TMP25]], ptr [[TMP76]], align 8
// CHECK1-NEXT: [[TMP78:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 9
// CHECK1-NEXT: store ptr [[TMP27]], ptr [[TMP78]], align 8
// CHECK1-NEXT: [[TMP80:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 9
// CHECK1-NEXT: store ptr null, ptr [[TMP80]], align 8
// CHECK1-NEXT: [[TMP81:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 10
// CHECK1-NEXT: store ptr [[TMP28]], ptr [[TMP81]], align 8
// CHECK1-NEXT: [[TMP83:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 10
// CHECK1-NEXT: store ptr [[TMP30]], ptr [[TMP83]], align 8
// CHECK1-NEXT: [[TMP85:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 10
// CHECK1-NEXT: store ptr null, ptr [[TMP85]], align 8
// CHECK1-NEXT: [[TMP86:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP87:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP88:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP66:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP88]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP66]], align 4
// CHECK1-NEXT: [[TMP89:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP67:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 11, ptr [[TMP89]], align 4 // CHECK1-NEXT: store i32 11, ptr [[TMP67]], align 4
// CHECK1-NEXT: [[TMP90:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP68:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP86]], ptr [[TMP90]], align 8 // CHECK1-NEXT: store ptr [[TMP64]], ptr [[TMP68]], align 8
// CHECK1-NEXT: [[TMP91:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP69:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP87]], ptr [[TMP91]], align 8 // CHECK1-NEXT: store ptr [[TMP65]], ptr [[TMP69]], align 8
// CHECK1-NEXT: [[TMP92:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP70:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP92]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP70]], align 8
// CHECK1-NEXT: [[TMP93:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP71:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP93]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP71]], align 8
// CHECK1-NEXT: [[TMP94:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP72:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP94]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP72]], align 8
// CHECK1-NEXT: [[TMP95:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP73:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP95]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP73]], align 8
// CHECK1-NEXT: [[TMP96:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP74:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 0, ptr [[TMP96]], align 8 // CHECK1-NEXT: store i64 0, ptr [[TMP74]], align 8
// CHECK1-NEXT: [[TMP97:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l41.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP75:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l41.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP98:%.*]] = icmp ne i32 [[TMP97]], 0 // CHECK1-NEXT: [[TMP76:%.*]] = icmp ne i32 [[TMP75]], 0
// CHECK1-NEXT: br i1 [[TMP98]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP76]], 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]+}}_main_l41(i64 [[TMP11]], ptr [[TMP12]], ptr [[TMP13]], ptr [[TMP14]], ptr [[A]], ptr [[TMP15]]) #[[ATTR4:[0-9]+]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l41(i64 [[TMP11]], ptr [[TMP12]], ptr [[TMP13]], ptr [[TMP14]], ptr [[A]], ptr [[TMP15]]) #[[ATTR4:[0-9]+]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: [[TMP99:%.*]] = load ptr, ptr [[B]], align 8 // CHECK1-NEXT: [[TMP77:%.*]] = load ptr, ptr [[B]], align 8
// CHECK1-NEXT: store ptr [[TMP99]], ptr [[_TMP4]], align 8 // CHECK1-NEXT: store ptr [[TMP77]], ptr [[_TMP4]], align 8
// CHECK1-NEXT: [[TMP100:%.*]] = load ptr, ptr [[C]], align 8 // CHECK1-NEXT: [[TMP78:%.*]] = load ptr, ptr [[C]], align 8
// CHECK1-NEXT: store ptr [[TMP100]], ptr [[_TMP5]], align 8 // CHECK1-NEXT: store ptr [[TMP78]], ptr [[_TMP5]], align 8
// CHECK1-NEXT: [[TMP101:%.*]] = load ptr, ptr [[L]], align 8 // CHECK1-NEXT: [[TMP79:%.*]] = load ptr, ptr [[L]], align 8
// CHECK1-NEXT: store ptr [[TMP101]], ptr [[_TMP6]], align 8 // CHECK1-NEXT: store ptr [[TMP79]], ptr [[_TMP6]], align 8
// CHECK1-NEXT: [[TMP102:%.*]] = load ptr, ptr [[_TMP4]], align 8 // CHECK1-NEXT: [[TMP80:%.*]] = load ptr, ptr [[_TMP4]], align 8
// CHECK1-NEXT: [[TMP103:%.*]] = load ptr, ptr [[_TMP5]], align 8 // CHECK1-NEXT: [[TMP81:%.*]] = load ptr, ptr [[_TMP5]], align 8
// CHECK1-NEXT: [[TMP104:%.*]] = load ptr, ptr [[D]], align 8 // CHECK1-NEXT: [[TMP82:%.*]] = load ptr, ptr [[D]], align 8
// CHECK1-NEXT: [[TMP105:%.*]] = load ptr, ptr [[_TMP6]], align 8 // CHECK1-NEXT: [[TMP83:%.*]] = load ptr, ptr [[_TMP6]], align 8
// CHECK1-NEXT: [[TMP106:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP105]], i32 0, i32 0 // CHECK1-NEXT: [[TMP84:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP83]], i32 0, i32 0
// CHECK1-NEXT: [[TMP107:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP105]], i32 0, i32 0 // CHECK1-NEXT: [[TMP85:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP83]], i32 0, i32 0
// CHECK1-NEXT: [[TMP108:%.*]] = load ptr, ptr [[TMP107]], align 8 // CHECK1-NEXT: [[TMP86:%.*]] = load ptr, ptr [[TMP85]], align 8
// CHECK1-NEXT: [[TMP109:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP105]], i32 0, i32 1 // CHECK1-NEXT: [[TMP87:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP83]], i32 0, i32 1
// CHECK1-NEXT: [[TMP110:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP105]], i32 0, i32 1 // CHECK1-NEXT: [[TMP88:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP83]], i32 0, i32 1
// CHECK1-NEXT: [[TMP111:%.*]] = load ptr, ptr [[TMP110]], align 8 // CHECK1-NEXT: [[TMP89:%.*]] = load ptr, ptr [[TMP88]], align 8
// CHECK1-NEXT: [[TMP112:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP105]], i32 0, i32 2 // CHECK1-NEXT: [[TMP90:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP83]], i32 0, i32 2
// CHECK1-NEXT: [[TMP113:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP105]], i32 0, i32 2 // CHECK1-NEXT: [[TMP91:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP83]], i32 0, i32 2
// CHECK1-NEXT: [[TMP114:%.*]] = load ptr, ptr [[TMP113]], align 8 // CHECK1-NEXT: [[TMP92:%.*]] = load ptr, ptr [[TMP91]], align 8
// CHECK1-NEXT: [[TMP115:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP105]], i32 0, i32 3 // CHECK1-NEXT: [[TMP93:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP83]], i32 0, i32 3
// CHECK1-NEXT: [[TMP116:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP105]], i32 0, i32 3 // CHECK1-NEXT: [[TMP94:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP83]], i32 0, i32 3
// CHECK1-NEXT: [[TMP117:%.*]] = load ptr, ptr [[TMP116]], align 8 // CHECK1-NEXT: [[TMP95:%.*]] = load ptr, ptr [[TMP94]], align 8
// CHECK1-NEXT: [[TMP118:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP105]], i32 0, i32 4 // CHECK1-NEXT: [[TMP96:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP83]], i32 0, i32 4
// CHECK1-NEXT: [[TMP119:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP105]], i32 0, i32 4 // CHECK1-NEXT: [[TMP97:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP83]], i32 0, i32 4
// CHECK1-NEXT: [[TMP120:%.*]] = load ptr, ptr [[TMP119]], align 8 // CHECK1-NEXT: [[TMP98:%.*]] = load ptr, ptr [[TMP97]], align 8
// CHECK1-NEXT: [[TMP121:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK1-NEXT: [[TMP99:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[ARGC_ADDR]], ptr [[TMP121]], align 8 // CHECK1-NEXT: store ptr [[ARGC_ADDR]], ptr [[TMP99]], align 8
// CHECK1-NEXT: [[TMP123:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK1-NEXT: [[TMP100:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[ARGC_ADDR]], ptr [[TMP123]], align 8 // CHECK1-NEXT: store ptr [[ARGC_ADDR]], ptr [[TMP100]], align 8
// CHECK1-NEXT: [[TMP125:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0 // CHECK1-NEXT: [[TMP101:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP101]], align 8
// CHECK1-NEXT: [[TMP102:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[TMP80]], ptr [[TMP102]], align 8
// CHECK1-NEXT: [[TMP103:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[TMP80]], ptr [[TMP103]], align 8
// CHECK1-NEXT: [[TMP104:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1
// CHECK1-NEXT: store ptr null, ptr [[TMP104]], align 8
// CHECK1-NEXT: [[TMP105:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP81]], ptr [[TMP105]], align 8
// CHECK1-NEXT: [[TMP106:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP81]], ptr [[TMP106]], align 8
// CHECK1-NEXT: [[TMP107:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 2
// CHECK1-NEXT: store ptr null, ptr [[TMP107]], align 8
// CHECK1-NEXT: [[TMP108:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP82]], ptr [[TMP108]], align 8
// CHECK1-NEXT: [[TMP109:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP82]], ptr [[TMP109]], align 8
// CHECK1-NEXT: [[TMP110:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 3
// CHECK1-NEXT: store ptr null, ptr [[TMP110]], align 8
// CHECK1-NEXT: [[TMP111:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 4
// CHECK1-NEXT: store ptr [[A]], ptr [[TMP111]], align 8
// CHECK1-NEXT: [[TMP112:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 4
// CHECK1-NEXT: store ptr [[A]], ptr [[TMP112]], align 8
// CHECK1-NEXT: [[TMP113:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 4
// CHECK1-NEXT: store ptr null, ptr [[TMP113]], align 8
// CHECK1-NEXT: [[TMP114:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 5
// CHECK1-NEXT: store ptr [[TMP83]], ptr [[TMP114]], align 8
// CHECK1-NEXT: [[TMP115:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 5
// CHECK1-NEXT: store ptr [[TMP83]], ptr [[TMP115]], align 8
// CHECK1-NEXT: [[TMP116:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 5
// CHECK1-NEXT: store ptr null, ptr [[TMP116]], align 8
// CHECK1-NEXT: [[TMP117:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 6
// CHECK1-NEXT: store ptr [[TMP84]], ptr [[TMP117]], align 8
// CHECK1-NEXT: [[TMP118:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 6
// CHECK1-NEXT: store ptr [[TMP86]], ptr [[TMP118]], align 8
// CHECK1-NEXT: [[TMP119:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 6
// CHECK1-NEXT: store ptr null, ptr [[TMP119]], align 8
// CHECK1-NEXT: [[TMP120:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 7
// CHECK1-NEXT: store ptr [[TMP87]], ptr [[TMP120]], align 8
// CHECK1-NEXT: [[TMP121:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 7
// CHECK1-NEXT: store ptr [[TMP89]], ptr [[TMP121]], align 8
// CHECK1-NEXT: [[TMP122:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 7
// CHECK1-NEXT: store ptr null, ptr [[TMP122]], align 8
// CHECK1-NEXT: [[TMP123:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 8
// CHECK1-NEXT: store ptr [[TMP90]], ptr [[TMP123]], align 8
// CHECK1-NEXT: [[TMP124:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 8
// CHECK1-NEXT: store ptr [[TMP92]], ptr [[TMP124]], align 8
// CHECK1-NEXT: [[TMP125:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 8
// CHECK1-NEXT: store ptr null, ptr [[TMP125]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP125]], align 8
// CHECK1-NEXT: [[TMP126:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 1 // CHECK1-NEXT: [[TMP126:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 9
// CHECK1-NEXT: store ptr [[TMP102]], ptr [[TMP126]], align 8 // CHECK1-NEXT: store ptr [[TMP93]], ptr [[TMP126]], align 8
// CHECK1-NEXT: [[TMP128:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 1 // CHECK1-NEXT: [[TMP127:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 9
// CHECK1-NEXT: store ptr [[TMP102]], ptr [[TMP128]], align 8 // CHECK1-NEXT: store ptr [[TMP95]], ptr [[TMP127]], align 8
// CHECK1-NEXT: [[TMP130:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 1 // CHECK1-NEXT: [[TMP128:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 9
// CHECK1-NEXT: store ptr null, ptr [[TMP130]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP128]], align 8
// CHECK1-NEXT: [[TMP131:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 2 // CHECK1-NEXT: [[TMP129:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 10
// CHECK1-NEXT: store ptr [[TMP103]], ptr [[TMP131]], align 8 // CHECK1-NEXT: store ptr [[TMP96]], ptr [[TMP129]], align 8
// CHECK1-NEXT: [[TMP133:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 2 // CHECK1-NEXT: [[TMP130:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 10
// CHECK1-NEXT: store ptr [[TMP103]], ptr [[TMP133]], align 8 // CHECK1-NEXT: store ptr [[TMP98]], ptr [[TMP130]], align 8
// CHECK1-NEXT: [[TMP135:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 2 // CHECK1-NEXT: [[TMP131:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 10
// CHECK1-NEXT: store ptr null, ptr [[TMP135]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP131]], align 8
// CHECK1-NEXT: [[TMP136:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 3 // CHECK1-NEXT: [[TMP132:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[TMP104]], ptr [[TMP136]], align 8 // CHECK1-NEXT: [[TMP133:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK1-NEXT: [[TMP138:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP104]], ptr [[TMP138]], align 8
// CHECK1-NEXT: [[TMP140:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 3
// CHECK1-NEXT: store ptr null, ptr [[TMP140]], align 8
// CHECK1-NEXT: [[TMP141:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 4
// CHECK1-NEXT: store ptr [[A]], ptr [[TMP141]], align 8
// CHECK1-NEXT: [[TMP143:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 4
// CHECK1-NEXT: store ptr [[A]], ptr [[TMP143]], align 8
// CHECK1-NEXT: [[TMP145:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 4
// CHECK1-NEXT: store ptr null, ptr [[TMP145]], align 8
// CHECK1-NEXT: [[TMP146:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 5
// CHECK1-NEXT: store ptr [[TMP105]], ptr [[TMP146]], align 8
// CHECK1-NEXT: [[TMP148:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 5
// CHECK1-NEXT: store ptr [[TMP105]], ptr [[TMP148]], align 8
// CHECK1-NEXT: [[TMP150:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 5
// CHECK1-NEXT: store ptr null, ptr [[TMP150]], align 8
// CHECK1-NEXT: [[TMP151:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 6
// CHECK1-NEXT: store ptr [[TMP106]], ptr [[TMP151]], align 8
// CHECK1-NEXT: [[TMP153:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 6
// CHECK1-NEXT: store ptr [[TMP108]], ptr [[TMP153]], align 8
// CHECK1-NEXT: [[TMP155:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 6
// CHECK1-NEXT: store ptr null, ptr [[TMP155]], align 8
// CHECK1-NEXT: [[TMP156:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 7
// CHECK1-NEXT: store ptr [[TMP109]], ptr [[TMP156]], align 8
// CHECK1-NEXT: [[TMP158:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 7
// CHECK1-NEXT: store ptr [[TMP111]], ptr [[TMP158]], align 8
// CHECK1-NEXT: [[TMP160:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 7
// CHECK1-NEXT: store ptr null, ptr [[TMP160]], align 8
// CHECK1-NEXT: [[TMP161:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 8
// CHECK1-NEXT: store ptr [[TMP112]], ptr [[TMP161]], align 8
// CHECK1-NEXT: [[TMP163:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 8
// CHECK1-NEXT: store ptr [[TMP114]], ptr [[TMP163]], align 8
// CHECK1-NEXT: [[TMP165:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 8
// CHECK1-NEXT: store ptr null, ptr [[TMP165]], align 8
// CHECK1-NEXT: [[TMP166:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 9
// CHECK1-NEXT: store ptr [[TMP115]], ptr [[TMP166]], align 8
// CHECK1-NEXT: [[TMP168:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 9
// CHECK1-NEXT: store ptr [[TMP117]], ptr [[TMP168]], align 8
// CHECK1-NEXT: [[TMP170:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 9
// CHECK1-NEXT: store ptr null, ptr [[TMP170]], align 8
// CHECK1-NEXT: [[TMP171:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 10
// CHECK1-NEXT: store ptr [[TMP118]], ptr [[TMP171]], align 8
// CHECK1-NEXT: [[TMP173:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 10
// CHECK1-NEXT: store ptr [[TMP120]], ptr [[TMP173]], align 8
// CHECK1-NEXT: [[TMP175:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 10
// CHECK1-NEXT: store ptr null, ptr [[TMP175]], align 8
// CHECK1-NEXT: [[TMP176:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK1-NEXT: [[TMP177:%.*]] = getelementptr inbounds [11 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS10:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS10:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK1-NEXT: [[TMP178:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 0 // CHECK1-NEXT: [[TMP134:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP178]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP134]], align 4
// CHECK1-NEXT: [[TMP179:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 1 // CHECK1-NEXT: [[TMP135:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 1
// CHECK1-NEXT: store i32 11, ptr [[TMP179]], align 4 // CHECK1-NEXT: store i32 11, ptr [[TMP135]], align 4
// CHECK1-NEXT: [[TMP180:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 2 // CHECK1-NEXT: [[TMP136:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP176]], ptr [[TMP180]], align 8 // CHECK1-NEXT: store ptr [[TMP132]], ptr [[TMP136]], align 8
// CHECK1-NEXT: [[TMP181:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 3 // CHECK1-NEXT: [[TMP137:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP177]], ptr [[TMP181]], align 8 // CHECK1-NEXT: store ptr [[TMP133]], ptr [[TMP137]], align 8
// CHECK1-NEXT: [[TMP182:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 4 // CHECK1-NEXT: [[TMP138:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.1, ptr [[TMP182]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.1, ptr [[TMP138]], align 8
// CHECK1-NEXT: [[TMP183:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 5 // CHECK1-NEXT: [[TMP139:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP183]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP139]], align 8
// CHECK1-NEXT: [[TMP184:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 6 // CHECK1-NEXT: [[TMP140:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP184]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP140]], align 8
// CHECK1-NEXT: [[TMP185:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 7 // CHECK1-NEXT: [[TMP141:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP185]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP141]], align 8
// CHECK1-NEXT: [[TMP186:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 8 // CHECK1-NEXT: [[TMP142:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 8
// CHECK1-NEXT: store i64 0, ptr [[TMP186]], align 8 // CHECK1-NEXT: store i64 0, ptr [[TMP142]], align 8
// CHECK1-NEXT: [[TMP187:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l43.region_id, ptr [[KERNEL_ARGS10]]) // CHECK1-NEXT: [[TMP143:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l43.region_id, ptr [[KERNEL_ARGS10]])
// CHECK1-NEXT: [[TMP188:%.*]] = icmp ne i32 [[TMP187]], 0 // CHECK1-NEXT: [[TMP144:%.*]] = icmp ne i32 [[TMP143]], 0
// CHECK1-NEXT: br i1 [[TMP188]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]] // CHECK1-NEXT: br i1 [[TMP144]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK1: omp_offload.failed11: // CHECK1: omp_offload.failed11:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l43(ptr [[ARGC_ADDR]], ptr [[TMP102]], ptr [[TMP103]], ptr [[TMP104]], ptr [[A]], ptr [[TMP105]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l43(ptr [[ARGC_ADDR]], ptr [[TMP80]], ptr [[TMP81]], ptr [[TMP82]], ptr [[A]], ptr [[TMP83]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK1: omp_offload.cont12: // CHECK1: omp_offload.cont12:
// CHECK1-NEXT: [[TMP189:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK1-NEXT: [[TMP145:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN1S3fooEv(ptr noundef nonnull align 4 dereferenceable(4) @s) // CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN1S3fooEv(ptr noundef nonnull align 4 dereferenceable(4) @s)
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP189]], [[CALL]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP145]], [[CALL]]
// CHECK1-NEXT: ret i32 [[ADD]] // CHECK1-NEXT: ret i32 [[ADD]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l41 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l41
// CHECK1-SAME: (i64 noundef [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (i64 noundef [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8
Show All 21 Lines
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[L_ADDR]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[L_ADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[_TMP1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[_TMP1]], align 8
// CHECK1-NEXT: store ptr [[TMP3]], ptr [[_TMP2]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L3]], ptr align 8 [[TMP4]], i64 40, i1 false) // CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L3]], ptr align 8 [[TMP4]], i64 40, i1 false)
// CHECK1-NEXT: store ptr [[L3]], ptr [[_TMP4]], align 8 // CHECK1-NEXT: store ptr [[L3]], ptr [[_TMP4]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: store i32 [[TMP8]], ptr [[B5]], align 4 // CHECK1-NEXT: store i32 [[TMP6]], ptr [[B5]], align 4
// CHECK1-NEXT: store ptr [[B5]], ptr [[_TMP6]], align 8 // CHECK1-NEXT: store ptr [[B5]], ptr [[_TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[_TMP1]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: store i32 [[TMP10]], ptr [[C7]], align 4 // CHECK1-NEXT: store i32 [[TMP8]], ptr [[C7]], align 4
// CHECK1-NEXT: store ptr [[C7]], ptr [[_TMP8]], align 8 // CHECK1-NEXT: store ptr [[C7]], ptr [[_TMP8]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[_TMP4]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[_TMP4]], align 8
// CHECK1-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(ptr noundef nonnull align 8 dereferenceable(40) [[TMP11]]) // CHECK1-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(ptr noundef nonnull align 8 dereferenceable(40) [[TMP9]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l43 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l43
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR5:[0-9]+]] { // CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR5:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8
▲ Show 20 Lines • Show All 62 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[L_ADDR]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[L_ADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8
// CHECK1-NEXT: store ptr [[TMP2]], ptr [[_TMP1]], align 8 // CHECK1-NEXT: store ptr [[TMP2]], ptr [[_TMP1]], align 8
// CHECK1-NEXT: store ptr [[TMP4]], ptr [[_TMP2]], align 8 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L3]], ptr align 8 [[TMP5]], i64 40, i1 false) // CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L3]], ptr align 8 [[TMP5]], i64 40, i1 false)
// CHECK1-NEXT: store ptr [[L3]], ptr [[_TMP4]], align 8 // CHECK1-NEXT: store ptr [[L3]], ptr [[_TMP4]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: store i32 [[TMP8]], ptr [[ARGC5]], align 4 // CHECK1-NEXT: store i32 [[TMP6]], ptr [[ARGC5]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: store i32 [[TMP10]], ptr [[B6]], align 4 // CHECK1-NEXT: store i32 [[TMP8]], ptr [[B6]], align 4
// CHECK1-NEXT: store ptr [[B6]], ptr [[_TMP7]], align 8 // CHECK1-NEXT: store ptr [[B6]], ptr [[_TMP7]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[_TMP1]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
// CHECK1-NEXT: store i32 [[TMP12]], ptr [[C8]], align 4 // CHECK1-NEXT: store i32 [[TMP10]], ptr [[C8]], align 4
// CHECK1-NEXT: store ptr [[C8]], ptr [[_TMP9]], align 8 // CHECK1-NEXT: store ptr [[C8]], ptr [[_TMP9]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP3]], align 4 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK1-NEXT: store i32 [[TMP13]], ptr [[A10]], align 4 // CHECK1-NEXT: store i32 [[TMP11]], ptr [[A10]], align 4
// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[_TMP4]], align 8 // CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[_TMP4]], align 8
// CHECK1-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(ptr noundef nonnull align 8 dereferenceable(40) [[TMP14]]) // CHECK1-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(ptr noundef nonnull align 8 dereferenceable(40) [[TMP12]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_ZN1S3fooEv // CHECK1-LABEL: define {{[^@]+}}@_ZN1S3fooEv
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) #[[ATTR3:[0-9]+]] comdat align 2 { // CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) #[[ATTR3:[0-9]+]] comdat align 2 {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[L:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[L:%.*]] = alloca ptr, align 8
Show All 13 Lines
// CHECK1-NEXT: store ptr [[REF_TMP]], ptr [[L]], align 8 // CHECK1-NEXT: store ptr [[REF_TMP]], ptr [[L]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP2]], i32 0, i32 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP2]], i32 0, i32 0
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP2]], i32 0, i32 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP2]], i32 0, i32 0
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP5]], align 8 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP5]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP7]], align 8 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP15]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP17]], align 8 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: store ptr null, ptr [[TMP19]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP22]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP16]], align 4
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 3, ptr [[TMP23]], align 4 // CHECK1-NEXT: store i32 3, ptr [[TMP17]], align 4
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP20]], ptr [[TMP24]], align 8 // CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP18]], align 8
// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP21]], ptr [[TMP25]], align 8 // CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP19]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.3, ptr [[TMP26]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.3, ptr [[TMP20]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP27]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP21]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP28]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP22]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP29]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP23]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 0, ptr [[TMP30]], align 8 // CHECK1-NEXT: store i64 0, ptr [[TMP24]], align 8
// CHECK1-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l27.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l27.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK1-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK1-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP26]], 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]+}}__ZN1S3fooEv_l27(ptr [[THIS1]], ptr [[TMP2]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l27(ptr [[THIS1]], ptr [[TMP2]]) #[[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: [[TMP33:%.*]] = load ptr, ptr [[L]], align 8 // CHECK1-NEXT: [[TMP27:%.*]] = load ptr, ptr [[L]], align 8
// CHECK1-NEXT: store ptr [[TMP33]], ptr [[_TMP2]], align 8 // CHECK1-NEXT: store ptr [[TMP27]], ptr [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP34:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK1-NEXT: [[TMP28:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP34]], i32 0, i32 0 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP28]], i32 0, i32 0
// CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP34]], i32 0, i32 0 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP28]], i32 0, i32 0
// CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP37]], align 8 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP31]], align 8
// CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP39]], align 8 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP32]], align 8
// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 0 // CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP41]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP33]], align 8
// CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 1 // CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[TMP34]], ptr [[TMP42]], align 8 // CHECK1-NEXT: store ptr [[TMP28]], ptr [[TMP34]], align 8
// CHECK1-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 1 // CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[TMP34]], ptr [[TMP44]], align 8 // CHECK1-NEXT: store ptr [[TMP28]], ptr [[TMP35]], align 8
// CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 1 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 1
// CHECK1-NEXT: store ptr null, ptr [[TMP46]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP36]], align 8
// CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 2 // CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP35]], ptr [[TMP47]], align 8 // CHECK1-NEXT: store ptr [[TMP29]], ptr [[TMP37]], align 8
// CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 2 // CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP36]], ptr [[TMP49]], align 8 // CHECK1-NEXT: store ptr [[TMP30]], ptr [[TMP38]], align 8
// CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 2 // CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 2
// CHECK1-NEXT: store ptr null, ptr [[TMP51]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP39]], align 8
// CHECK1-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 // CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0
// CHECK1-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 // CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS6:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS6:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK1-NEXT: [[TMP54:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 0 // CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP54]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP42]], align 4
// CHECK1-NEXT: [[TMP55:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 1 // CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 1
// CHECK1-NEXT: store i32 3, ptr [[TMP55]], align 4 // CHECK1-NEXT: store i32 3, ptr [[TMP43]], align 4
// CHECK1-NEXT: [[TMP56:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 2 // CHECK1-NEXT: [[TMP44:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP52]], ptr [[TMP56]], align 8 // CHECK1-NEXT: store ptr [[TMP40]], ptr [[TMP44]], align 8
// CHECK1-NEXT: [[TMP57:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 3 // CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP53]], ptr [[TMP57]], align 8 // CHECK1-NEXT: store ptr [[TMP41]], ptr [[TMP45]], align 8
// CHECK1-NEXT: [[TMP58:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 4 // CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.6, ptr [[TMP58]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.6, ptr [[TMP46]], align 8
// CHECK1-NEXT: [[TMP59:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 5 // CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.7, ptr [[TMP59]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.7, ptr [[TMP47]], align 8
// CHECK1-NEXT: [[TMP60:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 6 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP60]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP48]], align 8
// CHECK1-NEXT: [[TMP61:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 7 // CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP61]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP49]], align 8
// CHECK1-NEXT: [[TMP62:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 8 // CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 8
// CHECK1-NEXT: store i64 0, ptr [[TMP62]], align 8 // CHECK1-NEXT: store i64 0, ptr [[TMP50]], align 8
// CHECK1-NEXT: [[TMP63:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l29.region_id, ptr [[KERNEL_ARGS6]]) // CHECK1-NEXT: [[TMP51:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l29.region_id, ptr [[KERNEL_ARGS6]])
// CHECK1-NEXT: [[TMP64:%.*]] = icmp ne i32 [[TMP63]], 0 // CHECK1-NEXT: [[TMP52:%.*]] = icmp ne i32 [[TMP51]], 0
// CHECK1-NEXT: br i1 [[TMP64]], label [[OMP_OFFLOAD_FAILED7:%.*]], label [[OMP_OFFLOAD_CONT8:%.*]] // CHECK1-NEXT: br i1 [[TMP52]], label [[OMP_OFFLOAD_FAILED7:%.*]], label [[OMP_OFFLOAD_CONT8:%.*]]
// CHECK1: omp_offload.failed7: // CHECK1: omp_offload.failed7:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l29(ptr [[THIS1]], ptr [[TMP34]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l29(ptr [[THIS1]], ptr [[TMP28]]) #[[ATTR4]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT8]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT8]]
// CHECK1: omp_offload.cont8: // CHECK1: omp_offload.cont8:
// CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], ptr [[THIS1]], i32 0, i32 0 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP65:%.*]] = load i32, ptr [[A]], align 4 // CHECK1-NEXT: [[TMP53:%.*]] = load i32, ptr [[A]], align 4
// CHECK1-NEXT: [[TMP66:%.*]] = load ptr, ptr [[L]], align 8 // CHECK1-NEXT: [[TMP54:%.*]] = load ptr, ptr [[L]], align 8
// CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z3fooIZN1S3fooEvEUlvE_EiRKT_(ptr noundef nonnull align 8 dereferenceable(8) [[TMP66]]) // CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z3fooIZN1S3fooEvEUlvE_EiRKT_(ptr noundef nonnull align 8 dereferenceable(8) [[TMP54]])
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP65]], [[CALL]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP53]], [[CALL]]
// CHECK1-NEXT: ret i32 [[ADD]] // CHECK1-NEXT: ret i32 [[ADD]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l27 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l27
// CHECK1-SAME: (ptr noundef [[THIS:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noundef [[THIS:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[L_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[L_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[TMP:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[L1:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8 // CHECK1-NEXT: [[L1:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK1-NEXT: [[_TMP2:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[_TMP2:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8 // CHECK1-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L_ADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L1]], ptr align 8 [[TMP2]], i64 8, i1 false) // CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L1]], ptr align 8 [[TMP2]], i64 8, i1 false)
// CHECK1-NEXT: store ptr [[L1]], ptr [[_TMP2]], align 8 // CHECK1-NEXT: store ptr [[L1]], ptr [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP5]]) // CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP3]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_ZZN1S3fooEvENKUlvE_clEv // CHECK1-LABEL: define {{[^@]+}}@_ZZN1S3fooEvENKUlvE_clEv
// CHECK1-SAME: (ptr noundef nonnull align 8 dereferenceable(8) [[THIS:%.*]]) #[[ATTR3]] comdat align 2 { // CHECK1-SAME: (ptr noundef nonnull align 8 dereferenceable(8) [[THIS:%.*]]) #[[ATTR3]] comdat align 2 {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
Show All 36 Lines
// CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8 // CHECK1-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L_ADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L1]], ptr align 8 [[TMP2]], i64 8, i1 false) // CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L1]], ptr align 8 [[TMP2]], i64 8, i1 false)
// CHECK1-NEXT: store ptr [[L1]], ptr [[_TMP2]], align 8 // CHECK1-NEXT: store ptr [[L1]], ptr [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP5]]) // CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP3]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_Z3fooIZN1S3fooEvEUlvE_EiRKT_ // CHECK1-LABEL: define {{[^@]+}}@_Z3fooIZN1S3fooEvEUlvE_EiRKT_
// CHECK1-SAME: (ptr noundef nonnull align 8 dereferenceable(8) [[T:%.*]]) #[[ATTR3]] comdat { // CHECK1-SAME: (ptr noundef nonnull align 8 dereferenceable(8) [[T:%.*]]) #[[ATTR3]] comdat {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[T_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[T_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[TMP:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT: store ptr [[T]], ptr [[T_ADDR]], align 8 // CHECK1-NEXT: store ptr [[T]], ptr [[T_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[T_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[T_ADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[CLASS_ANON_0:%.*]], ptr [[TMP1]], i32 0, i32 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[CLASS_ANON_0:%.*]], ptr [[TMP1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP1]], i32 0, i32 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP6]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP5]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP8]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP11]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP16]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP12]], align 4
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 2, ptr [[TMP17]], align 4 // CHECK1-NEXT: store i32 2, ptr [[TMP13]], align 4
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP18]], align 8 // CHECK1-NEXT: store ptr [[TMP10]], ptr [[TMP14]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP19]], align 8 // CHECK1-NEXT: store ptr [[TMP11]], ptr [[TMP15]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.9, ptr [[TMP20]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.9, ptr [[TMP16]], align 8
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.10, ptr [[TMP21]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.10, ptr [[TMP17]], align 8
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP22]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP18]], align 8
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP23]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP19]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 0, ptr [[TMP24]], align 8 // CHECK1-NEXT: store i64 0, ptr [[TMP20]], align 8
// CHECK1-NEXT: [[TMP25:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooIZN1S3fooEvEUlvE_EiRKT__l18.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP21:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooIZN1S3fooEvEUlvE_EiRKT__l18.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP26:%.*]] = icmp ne i32 [[TMP25]], 0 // CHECK1-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK1-NEXT: br i1 [[TMP26]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP22]], 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]+}}__Z3fooIZN1S3fooEvEUlvE_EiRKT__l18(ptr [[TMP1]]) #[[ATTR4]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooIZN1S3fooEvEUlvE_EiRKT__l18(ptr [[TMP1]]) #[[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: ret i32 0 // CHECK1-NEXT: ret i32 0
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooIZN1S3fooEvEUlvE_EiRKT__l18 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooIZN1S3fooEvEUlvE_EiRKT__l18
Show All 21 Lines
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store ptr [[T]], ptr [[T_ADDR]], align 8 // CHECK1-NEXT: store ptr [[T]], ptr [[T_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[T_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[T_ADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[T1]], ptr align 8 [[TMP1]], i64 8, i1 false) // CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[T1]], ptr align 8 [[TMP1]], i64 8, i1 false)
// CHECK1-NEXT: store ptr [[T1]], ptr [[_TMP2]], align 8 // CHECK1-NEXT: store ptr [[T1]], ptr [[_TMP2]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP4]]) // CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP2]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK1-SAME: () #[[ATTR6:[0-9]+]] { // CHECK1-SAME: () #[[ATTR6:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: call void @__tgt_register_requires(i64 1) // CHECK1-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l27 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l27
// CHECK2-SAME: (ptr noundef [[THIS:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK2-SAME: (ptr noundef [[THIS:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[L_ADDR:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[L_ADDR:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[TMP:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[TMP:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[L1:%.*]] = alloca [[CLASS_ANON:%.*]], align 8 // CHECK2-NEXT: [[L1:%.*]] = alloca [[CLASS_ANON:%.*]], align 8
// CHECK2-NEXT: [[_TMP2:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[_TMP2:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK2-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK2-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8 // CHECK2-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L_ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L_ADDR]], align 8
// CHECK2-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8 // CHECK2-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP2]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP2]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK2-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L1]], ptr align 8 [[TMP3]], i64 8, i1 false) // CHECK2-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L1]], ptr align 8 [[TMP3]], i64 8, i1 false)
// CHECK2-NEXT: store ptr [[L1]], ptr [[_TMP2]], align 8 // CHECK2-NEXT: store ptr [[L1]], ptr [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP4]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP5]], align 8
// CHECK2-NEXT: [[TMP6:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK2-NEXT: [[TMP6:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP6]], i32 0, i32 0 // CHECK2-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP6]]) #[[ATTR7:[0-9]+]]
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP7]], align 8 // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: [[TMP8:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK2-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP8]]) #[[ATTR7:[0-9]+]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@_ZZN1S3fooEvENKUlvE_clEv // CHECK2-LABEL: define {{[^@]+}}@_ZZN1S3fooEvENKUlvE_clEv
// CHECK2-SAME: (ptr noundef nonnull align 8 dereferenceable(8) [[THIS:%.*]]) #[[ATTR2:[0-9]+]] comdat align 2 { // CHECK2-SAME: (ptr noundef nonnull align 8 dereferenceable(8) [[THIS:%.*]]) #[[ATTR2:[0-9]+]] comdat align 2 {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 14 Lines
// CHECK2-NEXT: [[L_ADDR:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[L_ADDR:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[TMP:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[TMP:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [2 x ptr], align 8 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [2 x ptr], align 8
// CHECK2-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK2-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK2-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8 // CHECK2-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L_ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L_ADDR]], align 8
// CHECK2-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8 // CHECK2-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP2]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP2]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]]) // CHECK2-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK2-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK2-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP5]], align 8 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP5]], align 8
// CHECK2-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK2-NEXT: store ptr [[TMP4]], ptr [[TMP7]], align 8 // CHECK2-NEXT: store ptr [[TMP4]], ptr [[TMP6]], align 8
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP3]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP3]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2)
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR4:[0-9]+]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 9 Lines
// CHECK2-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK2-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK2-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8 // CHECK2-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L_ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L_ADDR]], align 8
// CHECK2-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8 // CHECK2-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK2-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L1]], ptr align 8 [[TMP2]], i64 8, i1 false) // CHECK2-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L1]], ptr align 8 [[TMP2]], i64 8, i1 false)
// CHECK2-NEXT: store ptr [[L1]], ptr [[_TMP2]], align 8 // CHECK2-NEXT: store ptr [[L1]], ptr [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP3]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP4]], align 8
// CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP5]], i32 0, i32 0 // CHECK2-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP5]]) #[[ATTR7]]
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP6]], align 8
// CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK2-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP7]]) #[[ATTR7]]
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l41 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l41
// CHECK2-SAME: (i64 noundef [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (i64 noundef [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR0]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8
Show All 18 Lines
// CHECK2-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8 // CHECK2-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[C_ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[C_ADDR]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[L_ADDR]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[L_ADDR]], align 8
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8
// CHECK2-NEXT: store ptr [[TMP1]], ptr [[_TMP1]], align 8 // CHECK2-NEXT: store ptr [[TMP1]], ptr [[_TMP1]], align 8
// CHECK2-NEXT: store ptr [[TMP3]], ptr [[_TMP2]], align 8 // CHECK2-NEXT: store ptr [[TMP3]], ptr [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP4]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP4]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK2-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L3]], ptr align 8 [[TMP5]], i64 40, i1 false) // CHECK2-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L3]], ptr align 8 [[TMP5]], i64 40, i1 false)
// CHECK2-NEXT: store ptr [[L3]], ptr [[_TMP4]], align 8 // CHECK2-NEXT: store ptr [[L3]], ptr [[_TMP4]], align 8
// CHECK2-NEXT: [[TMP8:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK2-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
// CHECK2-NEXT: store i32 [[TMP9]], ptr [[B5]], align 4 // CHECK2-NEXT: store i32 [[TMP7]], ptr [[B5]], align 4
// CHECK2-NEXT: store ptr [[B5]], ptr [[_TMP6]], align 8 // CHECK2-NEXT: store ptr [[B5]], ptr [[_TMP6]], align 8
// CHECK2-NEXT: [[TMP10:%.*]] = load ptr, ptr [[_TMP1]], align 8 // CHECK2-NEXT: [[TMP8:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
// CHECK2-NEXT: store i32 [[TMP11]], ptr [[C7]], align 4 // CHECK2-NEXT: store i32 [[TMP9]], ptr [[C7]], align 4
// CHECK2-NEXT: store ptr [[C7]], ptr [[_TMP8]], align 8 // CHECK2-NEXT: store ptr [[C7]], ptr [[_TMP8]], align 8
// CHECK2-NEXT: [[TMP12:%.*]] = load ptr, ptr [[_TMP4]], align 8 // CHECK2-NEXT: [[TMP10:%.*]] = load ptr, ptr [[_TMP4]], align 8
// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP12]], i32 0, i32 0 // CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP10]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[ARGC_ADDR]], ptr [[TMP13]], align 8 // CHECK2-NEXT: store ptr [[ARGC_ADDR]], ptr [[TMP11]], align 8
// CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP12]], i32 0, i32 1 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP10]], i32 0, i32 1
// CHECK2-NEXT: [[TMP15:%.*]] = load ptr, ptr [[_TMP6]], align 8 // CHECK2-NEXT: [[TMP13:%.*]] = load ptr, ptr [[_TMP6]], align 8
// CHECK2-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
// CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP10]], i32 0, i32 2
// CHECK2-NEXT: [[TMP15:%.*]] = load ptr, ptr [[_TMP8]], align 8
// CHECK2-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8 // CHECK2-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8
// CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP12]], i32 0, i32 2 // CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP10]], i32 0, i32 3
// CHECK2-NEXT: [[TMP17:%.*]] = load ptr, ptr [[_TMP8]], align 8 // CHECK2-NEXT: store ptr [[D_ADDR]], ptr [[TMP16]], align 8
// CHECK2-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8 // CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP10]], i32 0, i32 4
// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP12]], i32 0, i32 3 // CHECK2-NEXT: store ptr [[TMP2]], ptr [[TMP17]], align 8
// CHECK2-NEXT: store ptr [[D_ADDR]], ptr [[TMP18]], align 8 // CHECK2-NEXT: [[TMP18:%.*]] = load ptr, ptr [[_TMP4]], align 8
// CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP12]], i32 0, i32 4 // CHECK2-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(ptr noundef nonnull align 8 dereferenceable(40) [[TMP18]]) #[[ATTR7]]
// CHECK2-NEXT: store ptr [[TMP2]], ptr [[TMP19]], align 8 // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: [[TMP20:%.*]] = load ptr, ptr [[_TMP4]], align 8
// CHECK2-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(ptr noundef nonnull align 8 dereferenceable(40) [[TMP20]]) #[[ATTR7]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l43 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l43
// CHECK2-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR3]] { // CHECK2-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 16 Lines
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[ARGC_ADDR]], align 8 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[ARGC_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[B_ADDR]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[B_ADDR]], align 8
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[C_ADDR]], align 8 // CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[C_ADDR]], align 8
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[L_ADDR]], align 8 // CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[L_ADDR]], align 8
// CHECK2-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8 // CHECK2-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8
// CHECK2-NEXT: store ptr [[TMP2]], ptr [[_TMP1]], align 8 // CHECK2-NEXT: store ptr [[TMP2]], ptr [[_TMP1]], align 8
// CHECK2-NEXT: store ptr [[TMP4]], ptr [[_TMP2]], align 8 // CHECK2-NEXT: store ptr [[TMP4]], ptr [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP5:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP5:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP5]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP5]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2]]) // CHECK2-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK2-NEXT: [[TMP8:%.*]] = load ptr, ptr [[_TMP1]], align 8 // CHECK2-NEXT: [[TMP8:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK2-NEXT: [[TMP9:%.*]] = load ptr, ptr [[D_ADDR]], align 8 // CHECK2-NEXT: [[TMP9:%.*]] = load ptr, ptr [[D_ADDR]], align 8
// CHECK2-NEXT: [[TMP10:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK2-NEXT: [[TMP10:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP11]], align 8 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP11]], align 8
// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK2-NEXT: store ptr [[TMP7]], ptr [[TMP13]], align 8 // CHECK2-NEXT: store ptr [[TMP7]], ptr [[TMP12]], align 8
// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2 // CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK2-NEXT: store ptr [[TMP8]], ptr [[TMP15]], align 8 // CHECK2-NEXT: store ptr [[TMP8]], ptr [[TMP13]], align 8
// CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3 // CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK2-NEXT: store ptr [[TMP9]], ptr [[TMP17]], align 8 // CHECK2-NEXT: store ptr [[TMP9]], ptr [[TMP14]], align 8
// CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4 // CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
// CHECK2-NEXT: store ptr [[TMP3]], ptr [[TMP19]], align 8 // CHECK2-NEXT: store ptr [[TMP3]], ptr [[TMP15]], align 8
// CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 5 // CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 5
// CHECK2-NEXT: store ptr [[TMP10]], ptr [[TMP21]], align 8 // CHECK2-NEXT: store ptr [[TMP10]], ptr [[TMP16]], align 8
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP6]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 6) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP6]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 6)
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR4]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR4]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 30 Lines
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[L_ADDR]], align 8 // CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[L_ADDR]], align 8
// CHECK2-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8 // CHECK2-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8
// CHECK2-NEXT: store ptr [[TMP2]], ptr [[_TMP1]], align 8 // CHECK2-NEXT: store ptr [[TMP2]], ptr [[_TMP1]], align 8
// CHECK2-NEXT: store ptr [[TMP4]], ptr [[_TMP2]], align 8 // CHECK2-NEXT: store ptr [[TMP4]], ptr [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK2-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L3]], ptr align 8 [[TMP5]], i64 40, i1 false) // CHECK2-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L3]], ptr align 8 [[TMP5]], i64 40, i1 false)
// CHECK2-NEXT: store ptr [[L3]], ptr [[_TMP4]], align 8 // CHECK2-NEXT: store ptr [[L3]], ptr [[_TMP4]], align 8
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK2-NEXT: store i32 [[TMP8]], ptr [[ARGC5]], align 4 // CHECK2-NEXT: store i32 [[TMP6]], ptr [[ARGC5]], align 4
// CHECK2-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4 // CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK2-NEXT: store i32 [[TMP10]], ptr [[B6]], align 4 // CHECK2-NEXT: store i32 [[TMP8]], ptr [[B6]], align 4
// CHECK2-NEXT: store ptr [[B6]], ptr [[_TMP7]], align 8 // CHECK2-NEXT: store ptr [[B6]], ptr [[_TMP7]], align 8
// CHECK2-NEXT: [[TMP11:%.*]] = load ptr, ptr [[_TMP1]], align 8 // CHECK2-NEXT: [[TMP9:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK2-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4 // CHECK2-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
// CHECK2-NEXT: store i32 [[TMP12]], ptr [[C8]], align 4 // CHECK2-NEXT: store i32 [[TMP10]], ptr [[C8]], align 4
// CHECK2-NEXT: store ptr [[C8]], ptr [[_TMP9]], align 8 // CHECK2-NEXT: store ptr [[C8]], ptr [[_TMP9]], align 8
// CHECK2-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP3]], align 4 // CHECK2-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK2-NEXT: store i32 [[TMP13]], ptr [[A10]], align 4 // CHECK2-NEXT: store i32 [[TMP11]], ptr [[A10]], align 4
// CHECK2-NEXT: [[TMP14:%.*]] = load ptr, ptr [[_TMP4]], align 8 // CHECK2-NEXT: [[TMP12:%.*]] = load ptr, ptr [[_TMP4]], align 8
// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP14]], i32 0, i32 0 // CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP12]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[ARGC5]], ptr [[TMP15]], align 8 // CHECK2-NEXT: store ptr [[ARGC5]], ptr [[TMP13]], align 8
// CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP14]], i32 0, i32 1 // CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP12]], i32 0, i32 1
// CHECK2-NEXT: [[TMP17:%.*]] = load ptr, ptr [[_TMP7]], align 8 // CHECK2-NEXT: [[TMP15:%.*]] = load ptr, ptr [[_TMP7]], align 8
// CHECK2-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8
// CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP12]], i32 0, i32 2
// CHECK2-NEXT: [[TMP17:%.*]] = load ptr, ptr [[_TMP9]], align 8
// CHECK2-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8 // CHECK2-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8
// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP14]], i32 0, i32 2 // CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP12]], i32 0, i32 3
// CHECK2-NEXT: [[TMP19:%.*]] = load ptr, ptr [[_TMP9]], align 8 // CHECK2-NEXT: store ptr [[D_ADDR]], ptr [[TMP18]], align 8
// CHECK2-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8 // CHECK2-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP12]], i32 0, i32 4
// CHECK2-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP14]], i32 0, i32 3 // CHECK2-NEXT: store ptr [[A10]], ptr [[TMP19]], align 8
// CHECK2-NEXT: store ptr [[D_ADDR]], ptr [[TMP20]], align 8 // CHECK2-NEXT: [[TMP20:%.*]] = load ptr, ptr [[_TMP4]], align 8
// CHECK2-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP14]], i32 0, i32 4 // CHECK2-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(ptr noundef nonnull align 8 dereferenceable(40) [[TMP20]]) #[[ATTR7]]
// CHECK2-NEXT: store ptr [[A10]], ptr [[TMP21]], align 8
// CHECK2-NEXT: [[TMP22:%.*]] = load ptr, ptr [[_TMP4]], align 8
// CHECK2-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(ptr noundef nonnull align 8 dereferenceable(40) [[TMP22]]) #[[ATTR7]]
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooIZN1S3fooEvEUlvE_EiRKT__l18 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooIZN1S3fooEvEUlvE_EiRKT__l18
// CHECK2-SAME: (ptr noundef nonnull align 8 dereferenceable(8) [[T:%.*]]) #[[ATTR3]] { // CHECK2-SAME: (ptr noundef nonnull align 8 dereferenceable(8) [[T:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[T_ADDR:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[T_ADDR:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[TMP:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[TMP:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8
// CHECK2-NEXT: store ptr [[T]], ptr [[T_ADDR]], align 8 // CHECK2-NEXT: store ptr [[T]], ptr [[T_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[T_ADDR]], align 8 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[T_ADDR]], align 8
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK2-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK2-NEXT: store ptr [[TMP3]], ptr [[TMP4]], align 8 // CHECK2-NEXT: store ptr [[TMP3]], ptr [[TMP4]], align 8
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__2, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__2, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1)
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__2 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__2
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[T:%.*]]) #[[ATTR4]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[T:%.*]]) #[[ATTR4]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[T_ADDR:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[T_ADDR:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[TMP:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[TMP:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[T1:%.*]] = alloca [[CLASS_ANON:%.*]], align 8 // CHECK2-NEXT: [[T1:%.*]] = alloca [[CLASS_ANON:%.*]], align 8
// CHECK2-NEXT: [[_TMP2:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[_TMP2:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 // CHECK2-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK2-NEXT: store ptr [[T]], ptr [[T_ADDR]], align 8 // CHECK2-NEXT: store ptr [[T]], ptr [[T_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[T_ADDR]], align 8 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[T_ADDR]], align 8
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK2-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[T1]], ptr align 8 [[TMP1]], i64 8, i1 false) // CHECK2-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[T1]], ptr align 8 [[TMP1]], i64 8, i1 false)
// CHECK2-NEXT: store ptr [[T1]], ptr [[_TMP2]], align 8 // CHECK2-NEXT: store ptr [[T1]], ptr [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK2-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP5]]) #[[ATTR7]] // CHECK2-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP3]]) #[[ATTR7]]
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l41 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l41
// CHECK3-SAME: (i64 noundef [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK3-SAME: (i64 noundef [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8 // CHECK3-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8
Show All 18 Lines
// CHECK3-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8 // CHECK3-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[C_ADDR]], align 8 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[C_ADDR]], align 8
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[L_ADDR]], align 8 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[L_ADDR]], align 8
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[_TMP1]], align 8 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[_TMP1]], align 8
// CHECK3-NEXT: store ptr [[TMP3]], ptr [[_TMP2]], align 8 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK3-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP4]], -1 // CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP4]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry: // CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L3]], ptr align 8 [[TMP5]], i64 40, i1 false) // CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L3]], ptr align 8 [[TMP5]], i64 40, i1 false)
// CHECK3-NEXT: store ptr [[L3]], ptr [[_TMP4]], align 8 // CHECK3-NEXT: store ptr [[L3]], ptr [[_TMP4]], align 8
// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4
// CHECK3-NEXT: store i32 [[TMP9]], ptr [[B5]], align 4 // CHECK3-NEXT: store i32 [[TMP7]], ptr [[B5]], align 4
// CHECK3-NEXT: store ptr [[B5]], ptr [[_TMP6]], align 8 // CHECK3-NEXT: store ptr [[B5]], ptr [[_TMP6]], align 8
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[_TMP1]], align 8 // CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
// CHECK3-NEXT: store i32 [[TMP11]], ptr [[C7]], align 4 // CHECK3-NEXT: store i32 [[TMP9]], ptr [[C7]], align 4
// CHECK3-NEXT: store ptr [[C7]], ptr [[_TMP8]], align 8 // CHECK3-NEXT: store ptr [[C7]], ptr [[_TMP8]], align 8
// CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[_TMP4]], align 8 // CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[_TMP4]], align 8
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP12]], i32 0, i32 0 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP10]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[ARGC_ADDR]], ptr [[TMP13]], align 8 // CHECK3-NEXT: store ptr [[ARGC_ADDR]], ptr [[TMP11]], align 8
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP12]], i32 0, i32 1 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP10]], i32 0, i32 1
// CHECK3-NEXT: [[TMP15:%.*]] = load ptr, ptr [[_TMP6]], align 8 // CHECK3-NEXT: [[TMP13:%.*]] = load ptr, ptr [[_TMP6]], align 8
// CHECK3-NEXT: store ptr [[TMP13]], ptr [[TMP12]], align 8
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP10]], i32 0, i32 2
// CHECK3-NEXT: [[TMP15:%.*]] = load ptr, ptr [[_TMP8]], align 8
// CHECK3-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8 // CHECK3-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP12]], i32 0, i32 2 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP10]], i32 0, i32 3
// CHECK3-NEXT: [[TMP17:%.*]] = load ptr, ptr [[_TMP8]], align 8 // CHECK3-NEXT: store ptr [[D_ADDR]], ptr [[TMP16]], align 8
// CHECK3-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8 // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP10]], i32 0, i32 4
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP12]], i32 0, i32 3 // CHECK3-NEXT: store ptr [[TMP2]], ptr [[TMP17]], align 8
// CHECK3-NEXT: store ptr [[D_ADDR]], ptr [[TMP18]], align 8 // CHECK3-NEXT: [[TMP18:%.*]] = load ptr, ptr [[_TMP4]], align 8
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP12]], i32 0, i32 4 // CHECK3-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(ptr noundef nonnull align 8 dereferenceable(40) [[TMP18]]) #[[ATTR7:[0-9]+]]
// CHECK3-NEXT: store ptr [[TMP2]], ptr [[TMP19]], align 8 // CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK3-NEXT: [[TMP20:%.*]] = load ptr, ptr [[_TMP4]], align 8
// CHECK3-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(ptr noundef nonnull align 8 dereferenceable(40) [[TMP20]]) #[[ATTR7:[0-9]+]]
// CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// CHECK3: worker.exit: // CHECK3: worker.exit:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l43 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l43
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR3:[0-9]+]] { // CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
Show All 16 Lines
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[ARGC_ADDR]], align 8 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[ARGC_ADDR]], align 8
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[B_ADDR]], align 8 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[B_ADDR]], align 8
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[C_ADDR]], align 8 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[C_ADDR]], align 8
// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[L_ADDR]], align 8 // CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[L_ADDR]], align 8
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8
// CHECK3-NEXT: store ptr [[TMP2]], ptr [[_TMP1]], align 8 // CHECK3-NEXT: store ptr [[TMP2]], ptr [[_TMP1]], align 8
// CHECK3-NEXT: store ptr [[TMP4]], ptr [[_TMP2]], align 8 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP5:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK3-NEXT: [[TMP5:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP5]], -1 // CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP5]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry: // CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]]) // CHECK3-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[_TMP1]], align 8 // CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[D_ADDR]], align 8 // CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[D_ADDR]], align 8
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP11]], align 8 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP11]], align 8
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP13]], align 8 // CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP12]], align 8
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK3-NEXT: store ptr [[TMP8]], ptr [[TMP15]], align 8 // CHECK3-NEXT: store ptr [[TMP8]], ptr [[TMP13]], align 8
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP17]], align 8 // CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP14]], align 8
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP19]], align 8 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP15]], align 8
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 5 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [6 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 5
// CHECK3-NEXT: store ptr [[TMP10]], ptr [[TMP21]], align 8 // CHECK3-NEXT: store ptr [[TMP10]], ptr [[TMP16]], align 8
// CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP6]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 6) // CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP6]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 6)
// CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// CHECK3: worker.exit: // CHECK3: worker.exit:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR4:[0-9]+]] { // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]], ptr noundef [[D:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 8 dereferenceable(40) [[L:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
Show All 30 Lines
// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[L_ADDR]], align 8 // CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[L_ADDR]], align 8
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8
// CHECK3-NEXT: store ptr [[TMP2]], ptr [[_TMP1]], align 8 // CHECK3-NEXT: store ptr [[TMP2]], ptr [[_TMP1]], align 8
// CHECK3-NEXT: store ptr [[TMP4]], ptr [[_TMP2]], align 8 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L3]], ptr align 8 [[TMP5]], i64 40, i1 false) // CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L3]], ptr align 8 [[TMP5]], i64 40, i1 false)
// CHECK3-NEXT: store ptr [[L3]], ptr [[_TMP4]], align 8 // CHECK3-NEXT: store ptr [[L3]], ptr [[_TMP4]], align 8
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT: store i32 [[TMP8]], ptr [[ARGC5]], align 4 // CHECK3-NEXT: store i32 [[TMP6]], ptr [[ARGC5]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK3-NEXT: store i32 [[TMP10]], ptr [[B6]], align 4 // CHECK3-NEXT: store i32 [[TMP8]], ptr [[B6]], align 4
// CHECK3-NEXT: store ptr [[B6]], ptr [[_TMP7]], align 8 // CHECK3-NEXT: store ptr [[B6]], ptr [[_TMP7]], align 8
// CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[_TMP1]], align 8 // CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4
// CHECK3-NEXT: store i32 [[TMP12]], ptr [[C8]], align 4 // CHECK3-NEXT: store i32 [[TMP10]], ptr [[C8]], align 4
// CHECK3-NEXT: store ptr [[C8]], ptr [[_TMP9]], align 8 // CHECK3-NEXT: store ptr [[C8]], ptr [[_TMP9]], align 8
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP3]], align 4 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK3-NEXT: store i32 [[TMP13]], ptr [[A10]], align 4 // CHECK3-NEXT: store i32 [[TMP11]], ptr [[A10]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = load ptr, ptr [[_TMP4]], align 8 // CHECK3-NEXT: [[TMP12:%.*]] = load ptr, ptr [[_TMP4]], align 8
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP14]], i32 0, i32 0 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP12]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[ARGC5]], ptr [[TMP15]], align 8 // CHECK3-NEXT: store ptr [[ARGC5]], ptr [[TMP13]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP14]], i32 0, i32 1 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP12]], i32 0, i32 1
// CHECK3-NEXT: [[TMP17:%.*]] = load ptr, ptr [[_TMP7]], align 8 // CHECK3-NEXT: [[TMP15:%.*]] = load ptr, ptr [[_TMP7]], align 8
// CHECK3-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP12]], i32 0, i32 2
// CHECK3-NEXT: [[TMP17:%.*]] = load ptr, ptr [[_TMP9]], align 8
// CHECK3-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8 // CHECK3-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 8
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP14]], i32 0, i32 2 // CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP12]], i32 0, i32 3
// CHECK3-NEXT: [[TMP19:%.*]] = load ptr, ptr [[_TMP9]], align 8 // CHECK3-NEXT: store ptr [[D_ADDR]], ptr [[TMP18]], align 8
// CHECK3-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8 // CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP12]], i32 0, i32 4
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP14]], i32 0, i32 3 // CHECK3-NEXT: store ptr [[A10]], ptr [[TMP19]], align 8
// CHECK3-NEXT: store ptr [[D_ADDR]], ptr [[TMP20]], align 8 // CHECK3-NEXT: [[TMP20:%.*]] = load ptr, ptr [[_TMP4]], align 8
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[CLASS_ANON]], ptr [[TMP14]], i32 0, i32 4 // CHECK3-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(ptr noundef nonnull align 8 dereferenceable(40) [[TMP20]]) #[[ATTR7]]
// CHECK3-NEXT: store ptr [[A10]], ptr [[TMP21]], align 8
// CHECK3-NEXT: [[TMP22:%.*]] = load ptr, ptr [[_TMP4]], align 8
// CHECK3-NEXT: [[CALL:%.*]] = call noundef i64 @"_ZZ4mainENK3$_0clEv"(ptr noundef nonnull align 8 dereferenceable(40) [[TMP22]]) #[[ATTR7]]
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l27 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN1S3fooEv_l27
// CHECK3-SAME: (ptr noundef [[THIS:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR0]] { // CHECK3-SAME: (ptr noundef [[THIS:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR0]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK3-NEXT: [[L_ADDR:%.*]] = alloca ptr, align 8 // CHECK3-NEXT: [[L_ADDR:%.*]] = alloca ptr, align 8
// CHECK3-NEXT: [[TMP:%.*]] = alloca ptr, align 8 // CHECK3-NEXT: [[TMP:%.*]] = alloca ptr, align 8
// CHECK3-NEXT: [[L1:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8 // CHECK3-NEXT: [[L1:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK3-NEXT: [[_TMP2:%.*]] = alloca ptr, align 8 // CHECK3-NEXT: [[_TMP2:%.*]] = alloca ptr, align 8
// CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK3-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8 // CHECK3-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L_ADDR]], align 8 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L_ADDR]], align 8
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8
// CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP2]], -1 // CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP2]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry: // CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L1]], ptr align 8 [[TMP3]], i64 8, i1 false) // CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L1]], ptr align 8 [[TMP3]], i64 8, i1 false)
// CHECK3-NEXT: store ptr [[L1]], ptr [[_TMP2]], align 8 // CHECK3-NEXT: store ptr [[L1]], ptr [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP4]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP5]], align 8
// CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP6]], i32 0, i32 0 // CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP6]]) #[[ATTR7]]
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP7]], align 8 // CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP8]]) #[[ATTR7]]
// CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// CHECK3: worker.exit: // CHECK3: worker.exit:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@_ZZN1S3fooEvENKUlvE_clEv // CHECK3-LABEL: define {{[^@]+}}@_ZZN1S3fooEvENKUlvE_clEv
// CHECK3-SAME: (ptr noundef nonnull align 8 dereferenceable(8) [[THIS:%.*]]) #[[ATTR2:[0-9]+]] comdat align 2 { // CHECK3-SAME: (ptr noundef nonnull align 8 dereferenceable(8) [[THIS:%.*]]) #[[ATTR2:[0-9]+]] comdat align 2 {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
Show All 14 Lines
// CHECK3-NEXT: [[L_ADDR:%.*]] = alloca ptr, align 8 // CHECK3-NEXT: [[L_ADDR:%.*]] = alloca ptr, align 8
// CHECK3-NEXT: [[TMP:%.*]] = alloca ptr, align 8 // CHECK3-NEXT: [[TMP:%.*]] = alloca ptr, align 8
// CHECK3-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [2 x ptr], align 8 // CHECK3-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [2 x ptr], align 8
// CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK3-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8 // CHECK3-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L_ADDR]], align 8 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L_ADDR]], align 8
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8
// CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP2]], -1 // CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP2]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry: // CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2]]) // CHECK3-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP5]], align 8 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP5]], align 8
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP7]], align 8 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP6]], align 8
// CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP3]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2) // CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP3]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2)
// CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// CHECK3: worker.exit: // CHECK3: worker.exit:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR4]] { // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[L:%.*]]) #[[ATTR4]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
Show All 9 Lines
// CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK3-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8 // CHECK3-NEXT: store ptr [[L]], ptr [[L_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L_ADDR]], align 8 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[L_ADDR]], align 8
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[TMP]], align 8
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L1]], ptr align 8 [[TMP2]], i64 8, i1 false) // CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[L1]], ptr align 8 [[TMP2]], i64 8, i1 false)
// CHECK3-NEXT: store ptr [[L1]], ptr [[_TMP2]], align 8 // CHECK3-NEXT: store ptr [[L1]], ptr [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP4]], align 8
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[TMP5]], i32 0, i32 0 // CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP5]]) #[[ATTR7]]
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP6]], align 8
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP7]]) #[[ATTR7]]
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooIZN1S3fooEvEUlvE_EiRKT__l18 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooIZN1S3fooEvEUlvE_EiRKT__l18
// CHECK3-SAME: (ptr noundef nonnull align 8 dereferenceable(8) [[T:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (ptr noundef nonnull align 8 dereferenceable(8) [[T:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[T_ADDR:%.*]] = alloca ptr, align 8 // CHECK3-NEXT: [[T_ADDR:%.*]] = alloca ptr, align 8
// CHECK3-NEXT: [[TMP:%.*]] = alloca ptr, align 8 // CHECK3-NEXT: [[TMP:%.*]] = alloca ptr, align 8
// CHECK3-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8 // CHECK3-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8
// CHECK3-NEXT: store ptr [[T]], ptr [[T_ADDR]], align 8 // CHECK3-NEXT: store ptr [[T]], ptr [[T_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[T_ADDR]], align 8 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[T_ADDR]], align 8
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8
// CHECK3-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK3-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry: // CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2]]) // CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP4]], align 8 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP4]], align 8
// CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__2, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1) // CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__2, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1)
// CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// CHECK3: worker.exit: // CHECK3: worker.exit:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__2 // CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__2
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[T:%.*]]) #[[ATTR4]] { // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[T:%.*]]) #[[ATTR4]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK3-NEXT: [[T_ADDR:%.*]] = alloca ptr, align 8 // CHECK3-NEXT: [[T_ADDR:%.*]] = alloca ptr, align 8
// CHECK3-NEXT: [[TMP:%.*]] = alloca ptr, align 8 // CHECK3-NEXT: [[TMP:%.*]] = alloca ptr, align 8
// CHECK3-NEXT: [[T1:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8 // CHECK3-NEXT: [[T1:%.*]] = alloca [[CLASS_ANON_0:%.*]], align 8
// CHECK3-NEXT: [[_TMP2:%.*]] = alloca ptr, align 8 // CHECK3-NEXT: [[_TMP2:%.*]] = alloca ptr, align 8
// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK3-NEXT: store ptr [[T]], ptr [[T_ADDR]], align 8 // CHECK3-NEXT: store ptr [[T]], ptr [[T_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[T_ADDR]], align 8 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[T_ADDR]], align 8
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP]], align 8
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[T1]], ptr align 8 [[TMP1]], i64 8, i1 false) // CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[T1]], ptr align 8 [[TMP1]], i64 8, i1 false)
// CHECK3-NEXT: store ptr [[T1]], ptr [[_TMP2]], align 8 // CHECK3-NEXT: store ptr [[T1]], ptr [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP2]], align 8 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP5]]) #[[ATTR7]] // CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_ZZN1S3fooEvENKUlvE_clEv(ptr noundef nonnull align 8 dereferenceable(8) [[TMP3]]) #[[ATTR7]]
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //

clang/test/OpenMP/nvptx_multi_target_parallel_codegen.cpp

Show All 23 Linesint main() {
{ use(); } { use(); }
} }
#endif #endif
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l21 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l21
// CHECK1-SAME: () #[[ATTR0:[0-9]+]] { // CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x ptr], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]]) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 0) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 0)
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: call void @_Z3usev() #[[ATTR8:[0-9]+]] // CHECK1-NEXT: call void @_Z3usev() #[[ATTR8:[0-9]+]]
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_Z3usev // CHECK1-LABEL: define {{[^@]+}}@_Z3usev
// CHECK1-SAME: () #[[ATTR2:[0-9]+]] { // CHECK1-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x ptr], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB3:[0-9]+]]) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB3]], i32 [[TMP0]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 0) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 0)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l23 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l23
// CHECK1-SAME: () #[[ATTR5:[0-9]+]] { // CHECK1-SAME: () #[[ATTR5:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: call void @_Z3usev() #[[ATTR8]] // CHECK1-NEXT: call void @_Z3usev() #[[ATTR8]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
Show All 19 Lines
// CHECK1-NEXT: call void @__omp_outlined__1(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]]) #[[ATTR3:[0-9]+]] // CHECK1-NEXT: call void @__omp_outlined__1(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]]) #[[ATTR3:[0-9]+]]
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l21 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l21
// CHECK2-SAME: () #[[ATTR0:[0-9]+]] { // CHECK2-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x ptr], align 4 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x ptr], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]]) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 0) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 0)
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK2-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 // CHECK2-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK2-NEXT: call void @_Z3usev() #[[ATTR8:[0-9]+]] // CHECK2-NEXT: call void @_Z3usev() #[[ATTR8:[0-9]+]]
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@_Z3usev // CHECK2-LABEL: define {{[^@]+}}@_Z3usev
// CHECK2-SAME: () #[[ATTR2:[0-9]+]] { // CHECK2-SAME: () #[[ATTR2:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x ptr], align 4 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x ptr], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB3:[0-9]+]]) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB3]], i32 [[TMP0]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 0) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 0)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l23 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l23
// CHECK2-SAME: () #[[ATTR5:[0-9]+]] { // CHECK2-SAME: () #[[ATTR5:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: call void @_Z3usev() #[[ATTR8]] // CHECK2-NEXT: call void @_Z3usev() #[[ATTR8]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 22 Lines

clang/test/OpenMP/nvptx_nested_parallel_codegen.cpp

Show All 34 Lines
#endif #endif
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l25 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l25
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 // CHECK1-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: call void @_Z3usePi(ptr noundef [[TMP0]]) #[[ATTR7:[0-9]+]] // CHECK1-NEXT: call void @_Z3usePi(ptr noundef [[TMP0]]) #[[ATTR7:[0-9]+]]
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP3]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP3]], align 8
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 2, i32 -1, ptr @__omp_outlined__, ptr @__omp_outlined___wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 1) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 2, i32 -1, ptr @__omp_outlined__, ptr @__omp_outlined___wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 1)
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_Z3usePi // CHECK1-LABEL: define {{[^@]+}}@_Z3usePi
// CHECK1-SAME: (ptr noundef [[C:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (ptr noundef [[C:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]]) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 // CHECK1-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[C_ADDR]], ptr [[TMP1]], align 8 // CHECK1-NEXT: store ptr [[C_ADDR]], ptr [[TMP1]], align 8
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP0]], i32 1, i32 2, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 1) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 2, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
Show All 14 Lines
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2 // CHECK1-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2
// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]]) // CHECK1-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]])
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP3]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 8
// CHECK1-NEXT: call void @__omp_outlined__(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP5]]) #[[ATTR4:[0-9]+]] // CHECK1-NEXT: call void @__omp_outlined__(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP4]]) #[[ATTR4:[0-9]+]]
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR2]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[C:%.*]]) #[[ATTR2]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
Show All 13 Lines
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[ATOMIC_TEMP1:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[ATOMIC_TEMP1:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 // CHECK1-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8
// CHECK1-NEXT: call void @__atomic_load(i64 noundef 4, ptr noundef [[TMP0]], ptr noundef [[ATOMIC_TEMP]], i32 noundef 0) #[[ATTR7]] // CHECK1-NEXT: call void @__atomic_load(i64 noundef 4, ptr noundef [[TMP0]], ptr noundef [[ATOMIC_TEMP]], i32 noundef 0) #[[ATTR7]]
// CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]] // CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]]
// CHECK1: atomic_cont: // CHECK1: atomic_cont:
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[ATOMIC_TEMP]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP3]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP1]], 1
// CHECK1-NEXT: store i32 [[ADD]], ptr [[ATOMIC_TEMP1]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[ATOMIC_TEMP1]], align 4
// CHECK1-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @__atomic_compare_exchange(i64 noundef 4, ptr noundef [[TMP0]], ptr noundef [[ATOMIC_TEMP]], ptr noundef [[ATOMIC_TEMP1]], i32 noundef 0, i32 noundef 0) #[[ATTR7]] // CHECK1-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @__atomic_compare_exchange(i64 noundef 4, ptr noundef [[TMP0]], ptr noundef [[ATOMIC_TEMP]], ptr noundef [[ATOMIC_TEMP1]], i32 noundef 0, i32 noundef 0) #[[ATTR7]]
// CHECK1-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]] // CHECK1-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]]
// CHECK1: atomic_exit: // CHECK1: atomic_exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1_wrapper // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1_wrapper
// CHECK1-SAME: (i16 noundef zeroext [[TMP0:%.*]], i32 noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (i16 noundef zeroext [[TMP0:%.*]], i32 noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i16, align 2 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i16, align 2
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2 // CHECK1-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2
// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]]) // CHECK1-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]])
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP3]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 8
// CHECK1-NEXT: call void @__omp_outlined__1(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP5]]) #[[ATTR4]] // CHECK1-NEXT: call void @__omp_outlined__1(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP4]]) #[[ATTR4]]
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l25 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l25
// CHECK2-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK2-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 4 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 4
// CHECK2-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 4 // CHECK2-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: call void @_Z3usePi(ptr noundef [[TMP0]]) #[[ATTR7:[0-9]+]] // CHECK2-NEXT: call void @_Z3usePi(ptr noundef [[TMP0]]) #[[ATTR7:[0-9]+]]
// CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP3]], align 4 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP3]], align 4
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 2, i32 -1, ptr @__omp_outlined__, ptr @__omp_outlined___wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 1) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 2, i32 -1, ptr @__omp_outlined__, ptr @__omp_outlined___wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 1)
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@_Z3usePi // CHECK2-LABEL: define {{[^@]+}}@_Z3usePi
// CHECK2-SAME: (ptr noundef [[C:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK2-SAME: (ptr noundef [[C:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 4 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]]) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 4 // CHECK2-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[C_ADDR]], ptr [[TMP1]], align 4 // CHECK2-NEXT: store ptr [[C_ADDR]], ptr [[TMP1]], align 4
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP0]], i32 1, i32 2, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 1) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 2, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
Show All 14 Lines
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2 // CHECK2-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2
// CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]]) // CHECK2-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]])
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i32 0 // CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i32 0
// CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP3]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 4
// CHECK2-NEXT: call void @__omp_outlined__(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP5]]) #[[ATTR4:[0-9]+]] // CHECK2-NEXT: call void @__omp_outlined__(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP4]]) #[[ATTR4:[0-9]+]]
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[C:%.*]]) #[[ATTR2]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
Show All 13 Lines
// CHECK2-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[ATOMIC_TEMP1:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[ATOMIC_TEMP1:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 4 // CHECK2-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 4
// CHECK2-NEXT: call void @__atomic_load(i32 noundef 4, ptr noundef [[TMP0]], ptr noundef [[ATOMIC_TEMP]], i32 noundef 0) #[[ATTR7]] // CHECK2-NEXT: call void @__atomic_load(i32 noundef 4, ptr noundef [[TMP0]], ptr noundef [[ATOMIC_TEMP]], i32 noundef 0) #[[ATTR7]]
// CHECK2-NEXT: br label [[ATOMIC_CONT:%.*]] // CHECK2-NEXT: br label [[ATOMIC_CONT:%.*]]
// CHECK2: atomic_cont: // CHECK2: atomic_cont:
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[ATOMIC_TEMP]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, ptr [[ATOMIC_TEMP]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP3]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP1]], 1
// CHECK2-NEXT: store i32 [[ADD]], ptr [[ATOMIC_TEMP1]], align 4 // CHECK2-NEXT: store i32 [[ADD]], ptr [[ATOMIC_TEMP1]], align 4
// CHECK2-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @__atomic_compare_exchange(i32 noundef 4, ptr noundef [[TMP0]], ptr noundef [[ATOMIC_TEMP]], ptr noundef [[ATOMIC_TEMP1]], i32 noundef 0, i32 noundef 0) #[[ATTR7]] // CHECK2-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @__atomic_compare_exchange(i32 noundef 4, ptr noundef [[TMP0]], ptr noundef [[ATOMIC_TEMP]], ptr noundef [[ATOMIC_TEMP1]], i32 noundef 0, i32 noundef 0) #[[ATTR7]]
// CHECK2-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]] // CHECK2-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]]
// CHECK2: atomic_exit: // CHECK2: atomic_exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1_wrapper // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1_wrapper
// CHECK2-SAME: (i16 noundef zeroext [[TMP0:%.*]], i32 noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK2-SAME: (i16 noundef zeroext [[TMP0:%.*]], i32 noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i16, align 2 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i16, align 2
// CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2 // CHECK2-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2
// CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]]) // CHECK2-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]])
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i32 0 // CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i32 0
// CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP3]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 4
// CHECK2-NEXT: call void @__omp_outlined__1(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP5]]) #[[ATTR4]] // CHECK2-NEXT: call void @__omp_outlined__1(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP4]]) #[[ATTR4]]
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //

clang/test/OpenMP/nvptx_parallel_codegen.cpp

Show First 20 Lines • Show All 75 Lines • ▼ Show 20 Lines
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l26 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l26
// CHECK1-SAME: (i64 noundef [[A:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK1-SAME: (i64 noundef [[A:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x ptr], align 8
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS1:%.*]] = alloca [0 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS1:%.*]] = alloca [0 x ptr], align 8
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS2:%.*]] = alloca [0 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS2:%.*]] = alloca [0 x ptr], align 8
// CHECK1-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8 // CHECK1-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr @__omp_outlined___wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 0) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr @__omp_outlined___wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 0)
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 0, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS1]], i64 0) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 0, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS1]], i64 0)
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__2, ptr @__omp_outlined__2_wrapper, ptr [[CAPTURED_VARS_ADDRS2]], i64 0) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__2, ptr @__omp_outlined__2_wrapper, ptr [[CAPTURED_VARS_ADDRS2]], i64 0)
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[A_ADDR]], align 4 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP5]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK1-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 83 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x ptr], align 8
// CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK1-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8 // CHECK1-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8
// CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8 // CHECK1-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 1000 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 1000
// CHECK1-NEXT: [[TMP4:%.*]] = zext i1 [[CMP]] to i32 // CHECK1-NEXT: [[TMP4:%.*]] = zext i1 [[CMP]] to i32
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 [[TMP4]], i32 -1, i32 -1, ptr @__omp_outlined__3, ptr @__omp_outlined__3_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 0) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 [[TMP4]], i32 -1, i32 -1, ptr @__omp_outlined__3, ptr @__omp_outlined__3_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 0)
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[A_ADDR]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP6]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP5]], 1
// CHECK1-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load i16, ptr [[AA_ADDR]], align 2 // CHECK1-NEXT: [[TMP6:%.*]] = load i16, ptr [[AA_ADDR]], align 2
// CHECK1-NEXT: [[CONV3:%.*]] = sext i16 [[TMP7]] to i32 // CHECK1-NEXT: [[CONV:%.*]] = sext i16 [[TMP6]] to i32
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], 1 // CHECK1-NEXT: [[ADD1:%.*]] = add nsw i32 [[CONV]], 1
// CHECK1-NEXT: [[CONV5:%.*]] = trunc i32 [[ADD4]] to i16 // CHECK1-NEXT: [[CONV2:%.*]] = trunc i32 [[ADD1]] to i16
// CHECK1-NEXT: store i16 [[CONV5]], ptr [[AA_ADDR]], align 2 // CHECK1-NEXT: store i16 [[CONV2]], ptr [[AA_ADDR]], align 2
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 2 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 2
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP8]], 1 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP7]], 1
// CHECK1-NEXT: store i32 [[ADD6]], ptr [[ARRAYIDX]], align 4 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__3 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__3
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
Show All 25 Lines
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l55 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l55
// CHECK1-SAME: (i64 noundef [[A:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (i64 noundef [[A:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8 // CHECK1-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[A_ADDR]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK1-NEXT: [[A1:%.*]] = call align 16 ptr @__kmpc_alloc_shared(i64 4) // CHECK1-NEXT: [[A1:%.*]] = call align 16 ptr @__kmpc_alloc_shared(i64 4)
// CHECK1-NEXT: store i32 [[TMP1]], ptr [[A1]], align 4 // CHECK1-NEXT: store i32 [[TMP1]], ptr [[A1]], align 4
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[A1]], ptr [[TMP3]], align 8 // CHECK1-NEXT: store ptr [[A1]], ptr [[TMP3]], align 8
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__4, ptr @__omp_outlined__4_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 1) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__4, ptr @__omp_outlined__4_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 1)
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[A1]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[A1]], align 4
// CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP6]], 1 // CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP4]], 1
// CHECK1-NEXT: store i32 [[INC]], ptr [[A1]], align 4 // CHECK1-NEXT: store i32 [[INC]], ptr [[A1]], align 4
// CHECK1-NEXT: call void @__kmpc_free_shared(ptr [[A1]], i64 4) // CHECK1-NEXT: call void @__kmpc_free_shared(ptr [[A1]], i64 4)
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__4 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__4
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2 // CHECK1-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2
// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]]) // CHECK1-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]])
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP3]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 8
// CHECK1-NEXT: call void @__omp_outlined__4(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP5]]) #[[ATTR3]] // CHECK1-NEXT: call void @__omp_outlined__4(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP4]]) #[[ATTR3]]
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l26 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l26
// CHECK2-SAME: (i32 noundef [[A:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK2-SAME: (i32 noundef [[A:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x ptr], align 4 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x ptr], align 4
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS1:%.*]] = alloca [0 x ptr], align 4 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS1:%.*]] = alloca [0 x ptr], align 4
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS2:%.*]] = alloca [0 x ptr], align 4 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS2:%.*]] = alloca [0 x ptr], align 4
// CHECK2-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 // CHECK2-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr @__omp_outlined___wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 0) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr @__omp_outlined___wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 0)
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 0, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS1]], i32 0) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 0, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS1]], i32 0)
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__2, ptr @__omp_outlined__2_wrapper, ptr [[CAPTURED_VARS_ADDRS2]], i32 0) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__2, ptr @__omp_outlined__2_wrapper, ptr [[CAPTURED_VARS_ADDRS2]], i32 0)
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[A_ADDR]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP5]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK2-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4 // CHECK2-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
▲ Show 20 Lines • Show All 83 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x ptr], align 4 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x ptr], align 4
// CHECK2-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 // CHECK2-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
// CHECK2-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 // CHECK2-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 4 // CHECK2-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 1000 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 1000
// CHECK2-NEXT: [[TMP4:%.*]] = zext i1 [[CMP]] to i32 // CHECK2-NEXT: [[TMP4:%.*]] = zext i1 [[CMP]] to i32
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 [[TMP4]], i32 -1, i32 -1, ptr @__omp_outlined__3, ptr @__omp_outlined__3_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 0) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 [[TMP4]], i32 -1, i32 -1, ptr @__omp_outlined__3, ptr @__omp_outlined__3_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 0)
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[A_ADDR]], align 4 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP6]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP5]], 1
// CHECK2-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4 // CHECK2-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP7:%.*]] = load i16, ptr [[AA_ADDR]], align 2 // CHECK2-NEXT: [[TMP6:%.*]] = load i16, ptr [[AA_ADDR]], align 2
// CHECK2-NEXT: [[CONV1:%.*]] = sext i16 [[TMP7]] to i32 // CHECK2-NEXT: [[CONV:%.*]] = sext i16 [[TMP6]] to i32
// CHECK2-NEXT: [[ADD2:%.*]] = add nsw i32 [[CONV1]], 1 // CHECK2-NEXT: [[ADD1:%.*]] = add nsw i32 [[CONV]], 1
// CHECK2-NEXT: [[CONV3:%.*]] = trunc i32 [[ADD2]] to i16 // CHECK2-NEXT: [[CONV2:%.*]] = trunc i32 [[ADD1]] to i16
// CHECK2-NEXT: store i16 [[CONV3]], ptr [[AA_ADDR]], align 2 // CHECK2-NEXT: store i16 [[CONV2]], ptr [[AA_ADDR]], align 2
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 2 // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 2
// CHECK2-NEXT: [[TMP8:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
// CHECK2-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP8]], 1 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP7]], 1
// CHECK2-NEXT: store i32 [[ADD4]], ptr [[ARRAYIDX]], align 4 // CHECK2-NEXT: store i32 [[ADD3]], ptr [[ARRAYIDX]], align 4
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__3 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__3
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 25 Lines
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l55 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l55
// CHECK2-SAME: (i32 noundef [[A:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (i32 noundef [[A:%.*]]) #[[ATTR0]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 4 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 4
// CHECK2-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 // CHECK2-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, ptr [[A_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[A1:%.*]] = call align 4 ptr @__kmpc_alloc_shared(i32 4) // CHECK2-NEXT: [[A1:%.*]] = call align 4 ptr @__kmpc_alloc_shared(i32 4)
// CHECK2-NEXT: store i32 [[TMP1]], ptr [[A1]], align 4 // CHECK2-NEXT: store i32 [[TMP1]], ptr [[A1]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[A1]], ptr [[TMP3]], align 4 // CHECK2-NEXT: store ptr [[A1]], ptr [[TMP3]], align 4
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__4, ptr @__omp_outlined__4_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 1) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__4, ptr @__omp_outlined__4_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 1)
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[A1]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[A1]], align 4
// CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP6]], 1 // CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP4]], 1
// CHECK2-NEXT: store i32 [[INC]], ptr [[A1]], align 4 // CHECK2-NEXT: store i32 [[INC]], ptr [[A1]], align 4
// CHECK2-NEXT: call void @__kmpc_free_shared(ptr [[A1]], i32 4) // CHECK2-NEXT: call void @__kmpc_free_shared(ptr [[A1]], i32 4)
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__4 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__4
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2 // CHECK2-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2
// CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]]) // CHECK2-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]])
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i32 0 // CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i32 0
// CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP3]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 4
// CHECK2-NEXT: call void @__omp_outlined__4(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP5]]) #[[ATTR2]] // CHECK2-NEXT: call void @__omp_outlined__4(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP4]]) #[[ATTR2]]
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //

clang/test/OpenMP/nvptx_parallel_for_codegen.cpp

Show All 35 Lines
// CHECK-SAME: (i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK-SAME: (i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK-NEXT: entry: // CHECK-NEXT: entry:
// CHECK-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8 // CHECK-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8
// CHECK-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [2 x ptr], align 8 // CHECK-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [2 x ptr], align 8
// CHECK-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8 // CHECK-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8
// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8 // CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8
// CHECK-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK: user_code.entry: // CHECK: user_code.entry:
// CHECK-NEXT: [[D:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 4) // CHECK-NEXT: [[D:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 4)
// CHECK-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK-NEXT: store i32 [[TMP3]], ptr [[D]], align 4 // CHECK-NEXT: store i32 [[TMP3]], ptr [[D]], align 4
// CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK-NEXT: store ptr [[TMP0]], ptr [[TMP4]], align 8 // CHECK-NEXT: store ptr [[TMP0]], ptr [[TMP4]], align 8
// CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 // CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK-NEXT: store ptr [[D]], ptr [[TMP6]], align 8 // CHECK-NEXT: store ptr [[D]], ptr [[TMP5]], align 8
// CHECK-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr @__omp_outlined___wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 2) // CHECK-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr @__omp_outlined___wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 2)
// CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 3 // CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 3
// CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 // CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
// CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], 1 // CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX]], align 4 // CHECK-NEXT: store i32 [[ADD]], ptr [[ARRAYIDX]], align 4
// CHECK-NEXT: call void @__kmpc_free_shared(ptr [[D]], i64 4) // CHECK-NEXT: call void @__kmpc_free_shared(ptr [[D]], i64 4)
// CHECK-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK-NEXT: ret void // CHECK-NEXT: ret void
// CHECK: worker.exit: // CHECK: worker.exit:
// CHECK-NEXT: ret void // CHECK-NEXT: ret void
// //
// //
// CHECK-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[D:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[D:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK-NEXT: entry: // CHECK-NEXT: entry:
▲ Show 20 Lines • Show All 92 Lines • ▼ Show 20 Lines
// CHECK-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8 // CHECK-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8
// CHECK-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2 // CHECK-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2
// CHECK-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]]) // CHECK-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]])
// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8 // CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8
// CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0
// CHECK-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP3]], align 8 // CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 8
// CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 1 // CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 1
// CHECK-NEXT: [[TMP8:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 8
// CHECK-NEXT: call void @__omp_outlined__(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP5]], ptr [[TMP8]]) #[[ATTR3:[0-9]+]] // CHECK-NEXT: call void @__omp_outlined__(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP4]], ptr [[TMP6]]) #[[ATTR3:[0-9]+]]
// CHECK-NEXT: ret void // CHECK-NEXT: ret void
// //

clang/test/OpenMP/nvptx_target_codegen.cpp

Show First 20 Lines • Show All 147 Lines • ▼ Show 20 Lines
// CHECK1-SAME: (ptr [[PTR1:%.*]], ptr nonnull align 8 dereferenceable(8) [[PTR2:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK1-SAME: (ptr [[PTR1:%.*]], ptr nonnull align 8 dereferenceable(8) [[PTR2:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[PTR1_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[PTR1_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[PTR2_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[PTR2_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [2 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT: store ptr [[PTR1]], ptr [[PTR1_ADDR]], align 8 // CHECK1-NEXT: store ptr [[PTR1]], ptr [[PTR1_ADDR]], align 8
// CHECK1-NEXT: store ptr [[PTR2]], ptr [[PTR2_ADDR]], align 8 // CHECK1-NEXT: store ptr [[PTR2]], ptr [[PTR2_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR2_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR2_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[PTR1_ADDR]], ptr [[TMP3]], align 8 // CHECK1-NEXT: store ptr [[PTR1_ADDR]], ptr [[TMP3]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP5]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP4]], align 8
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP2]], i32 1, i32 2, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 2, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 2)
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK1-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr nonnull align 8 dereferenceable(8) [[PTR1:%.*]], ptr nonnull align 8 dereferenceable(8) [[PTR2:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr nonnull align 8 dereferenceable(8) [[PTR1:%.*]], ptr nonnull align 8 dereferenceable(8) [[PTR2:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
Show All 12 Lines
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP0]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP0]], align 8
// CHECK1-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 // CHECK1-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l39 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l39
// CHECK1-SAME: () #[[ATTR4:[0-9]+]] { // CHECK1-SAME: () #[[ATTR4:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l47 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l47
// CHECK1-SAME: (i64 [[AA:%.*]]) #[[ATTR4]] { // CHECK1-SAME: (i64 [[AA:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP1:%.*]] = load i16, ptr [[AA_ADDR]], align 2 // CHECK1-NEXT: [[TMP1:%.*]] = load i16, ptr [[AA_ADDR]], align 2
// CHECK1-NEXT: [[CONV1:%.*]] = sext i16 [[TMP1]] to i32 // CHECK1-NEXT: [[CONV:%.*]] = sext i16 [[TMP1]] to i32
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV1]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], 1
// CHECK1-NEXT: [[CONV2:%.*]] = trunc i32 [[ADD]] to i16 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i32 [[ADD]] to i16
// CHECK1-NEXT: store i16 [[CONV2]], ptr [[AA_ADDR]], align 2 // CHECK1-NEXT: store i16 [[CONV1]], ptr [[AA_ADDR]], align 2
// CHECK1-NEXT: [[TMP2:%.*]] = load i16, ptr [[AA_ADDR]], align 2 // CHECK1-NEXT: [[TMP2:%.*]] = load i16, ptr [[AA_ADDR]], align 2
// CHECK1-NEXT: [[CONV3:%.*]] = sext i16 [[TMP2]] to i32 // CHECK1-NEXT: [[CONV2:%.*]] = sext i16 [[TMP2]] to i32
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], 2 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV2]], 2
// CHECK1-NEXT: [[CONV5:%.*]] = trunc i32 [[ADD4]] to i16 // CHECK1-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i16
// CHECK1-NEXT: store i16 [[CONV5]], ptr [[AA_ADDR]], align 2 // CHECK1-NEXT: store i16 [[CONV4]], ptr [[AA_ADDR]], align 2
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l53 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l53
// CHECK1-SAME: (i64 [[A:%.*]], ptr nonnull align 4 dereferenceable(40) [[B:%.*]], i64 [[VLA:%.*]], ptr nonnull align 4 dereferenceable(4) [[BN:%.*]], ptr nonnull align 8 dereferenceable(400) [[C:%.*]], i64 [[VLA1:%.*]], i64 [[VLA3:%.*]], ptr nonnull align 8 dereferenceable(8) [[CN:%.*]], ptr nonnull align 8 dereferenceable(16) [[D:%.*]]) #[[ATTR4]] { // CHECK1-SAME: (i64 [[A:%.*]], ptr nonnull align 4 dereferenceable(40) [[B:%.*]], i64 [[VLA:%.*]], ptr nonnull align 4 dereferenceable(4) [[BN:%.*]], ptr nonnull align 8 dereferenceable(400) [[C:%.*]], i64 [[VLA1:%.*]], i64 [[VLA3:%.*]], ptr nonnull align 8 dereferenceable(8) [[CN:%.*]], ptr nonnull align 8 dereferenceable(16) [[D:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
Show All 18 Lines
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[BN_ADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[BN_ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[C_ADDR]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load i64, ptr [[VLA_ADDR4]], align 8 // CHECK1-NEXT: [[TMP5:%.*]] = load i64, ptr [[VLA_ADDR4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[CN_ADDR]], align 8 // CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[CN_ADDR]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[D_ADDR]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[D_ADDR]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP8:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP8]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP8]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[A_ADDR]], align 4 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], 1
// CHECK1-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x float], ptr [[TMP0]], i64 0, i64 2 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x float], ptr [[TMP0]], i64 0, i64 2
// CHECK1-NEXT: [[TMP10:%.*]] = load float, ptr [[ARRAYIDX]], align 4 // CHECK1-NEXT: [[TMP10:%.*]] = load float, ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT: [[CONV5:%.*]] = fpext float [[TMP10]] to double // CHECK1-NEXT: [[CONV:%.*]] = fpext float [[TMP10]] to double
// CHECK1-NEXT: [[ADD6:%.*]] = fadd double [[CONV5]], 1.000000e+00 // CHECK1-NEXT: [[ADD5:%.*]] = fadd double [[CONV]], 1.000000e+00
// CHECK1-NEXT: [[CONV7:%.*]] = fptrunc double [[ADD6]] to float // CHECK1-NEXT: [[CONV6:%.*]] = fptrunc double [[ADD5]] to float
// CHECK1-NEXT: store float [[CONV7]], ptr [[ARRAYIDX]], align 4 // CHECK1-NEXT: store float [[CONV6]], ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds float, ptr [[TMP2]], i64 3 // CHECK1-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds float, ptr [[TMP2]], i64 3
// CHECK1-NEXT: [[TMP11:%.*]] = load float, ptr [[ARRAYIDX8]], align 4 // CHECK1-NEXT: [[TMP11:%.*]] = load float, ptr [[ARRAYIDX7]], align 4
// CHECK1-NEXT: [[CONV9:%.*]] = fpext float [[TMP11]] to double // CHECK1-NEXT: [[CONV8:%.*]] = fpext float [[TMP11]] to double
// CHECK1-NEXT: [[ADD10:%.*]] = fadd double [[CONV9]], 1.000000e+00 // CHECK1-NEXT: [[ADD9:%.*]] = fadd double [[CONV8]], 1.000000e+00
// CHECK1-NEXT: [[CONV11:%.*]] = fptrunc double [[ADD10]] to float // CHECK1-NEXT: [[CONV10:%.*]] = fptrunc double [[ADD9]] to float
// CHECK1-NEXT: store float [[CONV11]], ptr [[ARRAYIDX8]], align 4 // CHECK1-NEXT: store float [[CONV10]], ptr [[ARRAYIDX7]], align 4
// CHECK1-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds [5 x [10 x double]], ptr [[TMP3]], i64 0, i64 1 // CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [5 x [10 x double]], ptr [[TMP3]], i64 0, i64 1
// CHECK1-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds [10 x double], ptr [[ARRAYIDX12]], i64 0, i64 2 // CHECK1-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds [10 x double], ptr [[ARRAYIDX11]], i64 0, i64 2
// CHECK1-NEXT: [[TMP12:%.*]] = load double, ptr [[ARRAYIDX13]], align 8 // CHECK1-NEXT: [[TMP12:%.*]] = load double, ptr [[ARRAYIDX12]], align 8
// CHECK1-NEXT: [[ADD14:%.*]] = fadd double [[TMP12]], 1.000000e+00 // CHECK1-NEXT: [[ADD13:%.*]] = fadd double [[TMP12]], 1.000000e+00
// CHECK1-NEXT: store double [[ADD14]], ptr [[ARRAYIDX13]], align 8 // CHECK1-NEXT: store double [[ADD13]], ptr [[ARRAYIDX12]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = mul nsw i64 1, [[TMP5]] // CHECK1-NEXT: [[TMP13:%.*]] = mul nsw i64 1, [[TMP5]]
// CHECK1-NEXT: [[ARRAYIDX15:%.*]] = getelementptr inbounds double, ptr [[TMP6]], i64 [[TMP13]] // CHECK1-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds double, ptr [[TMP6]], i64 [[TMP13]]
// CHECK1-NEXT: [[ARRAYIDX16:%.*]] = getelementptr inbounds double, ptr [[ARRAYIDX15]], i64 3 // CHECK1-NEXT: [[ARRAYIDX15:%.*]] = getelementptr inbounds double, ptr [[ARRAYIDX14]], i64 3
// CHECK1-NEXT: [[TMP14:%.*]] = load double, ptr [[ARRAYIDX16]], align 8 // CHECK1-NEXT: [[TMP14:%.*]] = load double, ptr [[ARRAYIDX15]], align 8
// CHECK1-NEXT: [[ADD17:%.*]] = fadd double [[TMP14]], 1.000000e+00 // CHECK1-NEXT: [[ADD16:%.*]] = fadd double [[TMP14]], 1.000000e+00
// CHECK1-NEXT: store double [[ADD17]], ptr [[ARRAYIDX16]], align 8 // CHECK1-NEXT: store double [[ADD16]], ptr [[ARRAYIDX15]], align 8
// CHECK1-NEXT: [[X:%.*]] = getelementptr inbounds [[STRUCT_TT:%.*]], ptr [[TMP7]], i32 0, i32 0 // CHECK1-NEXT: [[X:%.*]] = getelementptr inbounds [[STRUCT_TT:%.*]], ptr [[TMP7]], i32 0, i32 0
// CHECK1-NEXT: [[TMP15:%.*]] = load i64, ptr [[X]], align 8 // CHECK1-NEXT: [[TMP15:%.*]] = load i64, ptr [[X]], align 8
// CHECK1-NEXT: [[ADD18:%.*]] = add nsw i64 [[TMP15]], 1 // CHECK1-NEXT: [[ADD17:%.*]] = add nsw i64 [[TMP15]], 1
// CHECK1-NEXT: store i64 [[ADD18]], ptr [[X]], align 8 // CHECK1-NEXT: store i64 [[ADD17]], ptr [[X]], align 8
// CHECK1-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_TT]], ptr [[TMP7]], i32 0, i32 1 // CHECK1-NEXT: [[Y:%.*]] = getelementptr inbounds [[STRUCT_TT]], ptr [[TMP7]], i32 0, i32 1
// CHECK1-NEXT: [[TMP16:%.*]] = load i8, ptr [[Y]], align 8 // CHECK1-NEXT: [[TMP16:%.*]] = load i8, ptr [[Y]], align 8
// CHECK1-NEXT: [[CONV19:%.*]] = sext i8 [[TMP16]] to i32 // CHECK1-NEXT: [[CONV18:%.*]] = sext i8 [[TMP16]] to i32
// CHECK1-NEXT: [[ADD20:%.*]] = add nsw i32 [[CONV19]], 1 // CHECK1-NEXT: [[ADD19:%.*]] = add nsw i32 [[CONV18]], 1
// CHECK1-NEXT: [[CONV21:%.*]] = trunc i32 [[ADD20]] to i8 // CHECK1-NEXT: [[CONV20:%.*]] = trunc i32 [[ADD19]] to i8
// CHECK1-NEXT: store i8 [[CONV21]], ptr [[Y]], align 8 // CHECK1-NEXT: store i8 [[CONV20]], ptr [[Y]], align 8
// CHECK1-NEXT: [[CALL:%.*]] = call nonnull align 8 dereferenceable(8) ptr @_ZN2TTIxcEixEi(ptr nonnull align 8 dereferenceable(16) [[TMP7]], i32 0) #[[ATTR10:[0-9]+]] // CHECK1-NEXT: [[CALL:%.*]] = call nonnull align 8 dereferenceable(8) ptr @_ZN2TTIxcEixEi(ptr nonnull align 8 dereferenceable(16) [[TMP7]], i32 0) #[[ATTR10:[0-9]+]]
// CHECK1-NEXT: [[TMP17:%.*]] = load i64, ptr [[CALL]], align 8 // CHECK1-NEXT: [[TMP17:%.*]] = load i64, ptr [[CALL]], align 8
// CHECK1-NEXT: [[ADD22:%.*]] = add nsw i64 [[TMP17]], 1 // CHECK1-NEXT: [[ADD21:%.*]] = add nsw i64 [[TMP17]], 1
// CHECK1-NEXT: store i64 [[ADD22]], ptr [[CALL]], align 8 // CHECK1-NEXT: store i64 [[ADD21]], ptr [[CALL]], align 8
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_ZN2TTIxcEixEi // CHECK1-LABEL: define {{[^@]+}}@_ZN2TTIxcEixEi
// CHECK1-SAME: (ptr nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32 [[I:%.*]]) #[[ATTR5:[0-9]+]] comdat align 2 { // CHECK1-SAME: (ptr nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32 [[I:%.*]]) #[[ATTR5:[0-9]+]] comdat align 2 {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
Show All 13 Lines
// 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 ptr, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8 // CHECK1-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8
// CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: store i64 [[AAA]], ptr [[AAA_ADDR]], align 8 // CHECK1-NEXT: store i64 [[AAA]], ptr [[AAA_ADDR]], align 8
// CHECK1-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8 // CHECK1-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[A_ADDR]], align 4 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP2]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK1-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT: [[TMP3:%.*]] = load i16, ptr [[AA_ADDR]], align 2 // CHECK1-NEXT: [[TMP3:%.*]] = load i16, ptr [[AA_ADDR]], align 2
// CHECK1-NEXT: [[CONV3:%.*]] = sext i16 [[TMP3]] to i32 // CHECK1-NEXT: [[CONV:%.*]] = sext i16 [[TMP3]] to i32
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], 1 // CHECK1-NEXT: [[ADD1:%.*]] = add nsw i32 [[CONV]], 1
// CHECK1-NEXT: [[CONV5:%.*]] = trunc i32 [[ADD4]] to i16 // CHECK1-NEXT: [[CONV2:%.*]] = trunc i32 [[ADD1]] to i16
// CHECK1-NEXT: store i16 [[CONV5]], ptr [[AA_ADDR]], align 2 // CHECK1-NEXT: store i16 [[CONV2]], ptr [[AA_ADDR]], align 2
// CHECK1-NEXT: [[TMP4:%.*]] = load i8, ptr [[AAA_ADDR]], align 1 // CHECK1-NEXT: [[TMP4:%.*]] = load i8, ptr [[AAA_ADDR]], align 1
// CHECK1-NEXT: [[CONV6:%.*]] = sext i8 [[TMP4]] to i32 // CHECK1-NEXT: [[CONV3:%.*]] = sext i8 [[TMP4]] to i32
// CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[CONV6]], 1 // CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], 1
// CHECK1-NEXT: [[CONV8:%.*]] = trunc i32 [[ADD7]] to i8 // CHECK1-NEXT: [[CONV5:%.*]] = trunc i32 [[ADD4]] to i8
// CHECK1-NEXT: store i8 [[CONV8]], ptr [[AAA_ADDR]], align 1 // CHECK1-NEXT: store i8 [[CONV5]], ptr [[AAA_ADDR]], align 1
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 2 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 2
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP5]], 1 // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP5]], 1
// CHECK1-NEXT: store i32 [[ADD9]], ptr [[ARRAYIDX]], align 4 // CHECK1-NEXT: store i32 [[ADD6]], ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l108 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l108
// CHECK1-SAME: (ptr [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], ptr nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR4]] { // CHECK1-SAME: (ptr [[THIS:%.*]], i64 [[B:%.*]], i64 [[VLA:%.*]], i64 [[VLA1:%.*]], ptr nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, 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
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: store i64 [[B]], ptr [[B_ADDR]], align 8 // CHECK1-NEXT: store i64 [[B]], ptr [[B_ADDR]], align 8
// CHECK1-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 // CHECK1-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK1-NEXT: store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8 // CHECK1-NEXT: store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8
// CHECK1-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 // CHECK1-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[C_ADDR]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP4]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP4]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[B_ADDR]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[B_ADDR]], align 4
// CHECK1-NEXT: [[CONV3:%.*]] = sitofp i32 [[TMP5]] to double // CHECK1-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP5]] to double
// CHECK1-NEXT: [[ADD:%.*]] = fadd double [[CONV3]], 1.500000e+00 // CHECK1-NEXT: [[ADD:%.*]] = fadd double [[CONV]], 1.500000e+00
// CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S1:%.*]], ptr [[TMP0]], i32 0, i32 0 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S1:%.*]], ptr [[TMP0]], i32 0, i32 0
// CHECK1-NEXT: store double [[ADD]], ptr [[A]], align 8 // CHECK1-NEXT: store double [[ADD]], ptr [[A]], align 8
// CHECK1-NEXT: [[A4:%.*]] = getelementptr inbounds [[STRUCT_S1]], ptr [[TMP0]], i32 0, i32 0 // CHECK1-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_S1]], ptr [[TMP0]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = load double, ptr [[A4]], align 8 // CHECK1-NEXT: [[TMP6:%.*]] = load double, ptr [[A3]], align 8
// CHECK1-NEXT: [[INC:%.*]] = fadd double [[TMP6]], 1.000000e+00 // CHECK1-NEXT: [[INC:%.*]] = fadd double [[TMP6]], 1.000000e+00
// CHECK1-NEXT: store double [[INC]], ptr [[A4]], align 8 // CHECK1-NEXT: store double [[INC]], ptr [[A3]], align 8
// CHECK1-NEXT: [[CONV5:%.*]] = fptosi double [[INC]] to i16 // CHECK1-NEXT: [[CONV4:%.*]] = fptosi double [[INC]] to i16
// CHECK1-NEXT: [[TMP7:%.*]] = mul nsw i64 1, [[TMP2]] // CHECK1-NEXT: [[TMP7:%.*]] = mul nsw i64 1, [[TMP2]]
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[TMP3]], i64 [[TMP7]] // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[TMP3]], i64 [[TMP7]]
// CHECK1-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i16, ptr [[ARRAYIDX]], i64 1 // CHECK1-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i16, ptr [[ARRAYIDX]], i64 1
// CHECK1-NEXT: store i16 [[CONV5]], ptr [[ARRAYIDX6]], align 2 // CHECK1-NEXT: store i16 [[CONV4]], ptr [[ARRAYIDX5]], align 2
// CHECK1-NEXT: [[A7:%.*]] = getelementptr inbounds [[STRUCT_S1]], ptr [[TMP0]], i32 0, i32 0 // CHECK1-NEXT: [[A6:%.*]] = getelementptr inbounds [[STRUCT_S1]], ptr [[TMP0]], i32 0, i32 0
// CHECK1-NEXT: [[TMP8:%.*]] = load double, ptr [[A7]], align 8 // CHECK1-NEXT: [[TMP8:%.*]] = load double, ptr [[A6]], align 8
// CHECK1-NEXT: [[CONV8:%.*]] = fptosi double [[TMP8]] to i32 // CHECK1-NEXT: [[CONV7:%.*]] = fptosi double [[TMP8]] to i32
// CHECK1-NEXT: [[A9:%.*]] = getelementptr inbounds [[STRUCT_S1]], ptr [[TMP0]], i32 0, i32 0 // CHECK1-NEXT: [[A8:%.*]] = getelementptr inbounds [[STRUCT_S1]], ptr [[TMP0]], i32 0, i32 0
// CHECK1-NEXT: [[CALL:%.*]] = call i32 @_Z3baziRd(i32 [[CONV8]], ptr nonnull align 8 dereferenceable(8) [[A9]]) #[[ATTR10]] // CHECK1-NEXT: [[CALL:%.*]] = call i32 @_Z3baziRd(i32 [[CONV7]], ptr nonnull align 8 dereferenceable(8) [[A8]]) #[[ATTR10]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_Z3baziRd // CHECK1-LABEL: define {{[^@]+}}@_Z3baziRd
// CHECK1-SAME: (i32 [[F1:%.*]], ptr nonnull align 8 dereferenceable(8) [[A:%.*]]) #[[ATTR5]] { // CHECK1-SAME: (i32 [[F1:%.*]], ptr nonnull align 8 dereferenceable(8) [[A:%.*]]) #[[ATTR5]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [2 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB3:[0-9]+]]) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[F:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 4) // CHECK1-NEXT: [[F:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 4)
// CHECK1-NEXT: store i32 [[F1]], ptr [[F]], align 4 // CHECK1-NEXT: store i32 [[F1]], ptr [[F]], align 4
// CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[F]], ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr [[F]], ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP3]], align 8
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB3]], i32 [[TMP0]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 2) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 2)
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[F]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[F]], align 4
// CHECK1-NEXT: call void @__kmpc_free_shared(ptr [[F]], i64 4) // CHECK1-NEXT: call void @__kmpc_free_shared(ptr [[F]], i64 4)
// CHECK1-NEXT: ret i32 [[TMP7]] // CHECK1-NEXT: ret i32 [[TMP4]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16unreachable_callv_l142 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16unreachable_callv_l142
// CHECK1-SAME: () #[[ATTR4]] { // CHECK1-SAME: () #[[ATTR4]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: call void @_Z6asserti(i32 0) #[[ATTR11:[0-9]+]] // CHECK1-NEXT: call void @_Z6asserti(i32 0) #[[ATTR11:[0-9]+]]
// CHECK1-NEXT: unreachable // CHECK1-NEXT: unreachable
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: 1: // CHECK1: 1:
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l74 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l74
// CHECK1-SAME: (i64 [[A:%.*]], i64 [[AA:%.*]], ptr nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR4]] { // CHECK1-SAME: (i64 [[A:%.*]], i64 [[AA:%.*]], ptr nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// 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 ptr, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8 // CHECK1-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8
// CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8 // CHECK1-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[A_ADDR]], align 4 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP2]], 1 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK1-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT: [[TMP3:%.*]] = load i16, ptr [[AA_ADDR]], align 2 // CHECK1-NEXT: [[TMP3:%.*]] = load i16, ptr [[AA_ADDR]], align 2
// CHECK1-NEXT: [[CONV2:%.*]] = sext i16 [[TMP3]] to i32 // CHECK1-NEXT: [[CONV:%.*]] = sext i16 [[TMP3]] to i32
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV2]], 1 // CHECK1-NEXT: [[ADD1:%.*]] = add nsw i32 [[CONV]], 1
// CHECK1-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i16 // CHECK1-NEXT: [[CONV2:%.*]] = trunc i32 [[ADD1]] to i16
// CHECK1-NEXT: store i16 [[CONV4]], ptr [[AA_ADDR]], align 2 // CHECK1-NEXT: store i16 [[CONV2]], ptr [[AA_ADDR]], align 2
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 2 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 2
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP4]], 1 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP4]], 1
// CHECK1-NEXT: store i32 [[ADD5]], ptr [[ARRAYIDX]], align 4 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK1-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr nonnull align 4 dereferenceable(4) [[F:%.*]], ptr nonnull align 8 dereferenceable(8) [[A:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr nonnull align 4 dereferenceable(4) [[F:%.*]], ptr nonnull align 8 dereferenceable(8) [[A:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
Show All 25 Lines
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2 // CHECK1-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2
// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]]) // CHECK1-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]])
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP3]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 1 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 1
// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 8
// CHECK1-NEXT: call void @__omp_outlined__1(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP5]], ptr [[TMP8]]) #[[ATTR2:[0-9]+]] // CHECK1-NEXT: call void @__omp_outlined__1(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP4]], ptr [[TMP6]]) #[[ATTR2:[0-9]+]]
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9targetBarPiS__l25 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9targetBarPiS__l25
// CHECK2-SAME: (ptr [[PTR1:%.*]], ptr nonnull align 4 dereferenceable(4) [[PTR2:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK2-SAME: (ptr [[PTR1:%.*]], ptr nonnull align 4 dereferenceable(4) [[PTR2:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[PTR1_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[PTR1_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[PTR2_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[PTR2_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [2 x ptr], align 4 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [2 x ptr], align 4
// CHECK2-NEXT: store ptr [[PTR1]], ptr [[PTR1_ADDR]], align 4 // CHECK2-NEXT: store ptr [[PTR1]], ptr [[PTR1_ADDR]], align 4
// CHECK2-NEXT: store ptr [[PTR2]], ptr [[PTR2_ADDR]], align 4 // CHECK2-NEXT: store ptr [[PTR2]], ptr [[PTR2_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR2_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[PTR2_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[PTR1_ADDR]], ptr [[TMP3]], align 4 // CHECK2-NEXT: store ptr [[PTR1_ADDR]], ptr [[TMP3]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1 // CHECK2-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP5]], align 4 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP4]], align 4
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP2]], i32 1, i32 2, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 2, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 2)
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK2-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr nonnull align 4 dereferenceable(4) [[PTR1:%.*]], ptr nonnull align 4 dereferenceable(4) [[PTR2:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK2-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr nonnull align 4 dereferenceable(4) [[PTR1:%.*]], ptr nonnull align 4 dereferenceable(4) [[PTR2:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 12 Lines
// CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP0]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP0]], align 4
// CHECK2-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4 // CHECK2-NEXT: store i32 [[TMP3]], ptr [[TMP4]], align 4
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l39 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l39
// CHECK2-SAME: () #[[ATTR4:[0-9]+]] { // CHECK2-SAME: () #[[ATTR4:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l47 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l47
// CHECK2-SAME: (i32 [[AA:%.*]]) #[[ATTR4]] { // CHECK2-SAME: (i32 [[AA:%.*]]) #[[ATTR4]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP1:%.*]] = load i16, ptr [[AA_ADDR]], align 2 // CHECK2-NEXT: [[TMP1:%.*]] = load i16, ptr [[AA_ADDR]], align 2
// CHECK2-NEXT: [[CONV1:%.*]] = sext i16 [[TMP1]] to i32 // CHECK2-NEXT: [[CONV:%.*]] = sext i16 [[TMP1]] to i32
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV1]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], 1
// CHECK2-NEXT: [[CONV2:%.*]] = trunc i32 [[ADD]] to i16 // CHECK2-NEXT: [[CONV1:%.*]] = trunc i32 [[ADD]] to i16
// CHECK2-NEXT: store i16 [[CONV2]], ptr [[AA_ADDR]], align 2 // CHECK2-NEXT: store i16 [[CONV1]], ptr [[AA_ADDR]], align 2
// CHECK2-NEXT: [[TMP2:%.*]] = load i16, ptr [[AA_ADDR]], align 2 // CHECK2-NEXT: [[TMP2:%.*]] = load i16, ptr [[AA_ADDR]], align 2
// CHECK2-NEXT: [[CONV3:%.*]] = sext i16 [[TMP2]] to i32 // CHECK2-NEXT: [[CONV2:%.*]] = sext i16 [[TMP2]] to i32
// CHECK2-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], 2 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV2]], 2
// CHECK2-NEXT: [[CONV5:%.*]] = trunc i32 [[ADD4]] to i16 // CHECK2-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i16
// CHECK2-NEXT: store i16 [[CONV5]], ptr [[AA_ADDR]], align 2 // CHECK2-NEXT: store i16 [[CONV4]], ptr [[AA_ADDR]], align 2
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l53 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l53
// CHECK2-SAME: (i32 [[A:%.*]], ptr nonnull align 4 dereferenceable(40) [[B:%.*]], i32 [[VLA:%.*]], ptr nonnull align 4 dereferenceable(4) [[BN:%.*]], ptr nonnull align 8 dereferenceable(400) [[C:%.*]], i32 [[VLA1:%.*]], i32 [[VLA3:%.*]], ptr nonnull align 8 dereferenceable(8) [[CN:%.*]], ptr nonnull align 8 dereferenceable(16) [[D:%.*]]) #[[ATTR4]] { // CHECK2-SAME: (i32 [[A:%.*]], ptr nonnull align 4 dereferenceable(40) [[B:%.*]], i32 [[VLA:%.*]], ptr nonnull align 4 dereferenceable(4) [[BN:%.*]], ptr nonnull align 8 dereferenceable(400) [[C:%.*]], i32 [[VLA1:%.*]], i32 [[VLA3:%.*]], ptr nonnull align 8 dereferenceable(8) [[CN:%.*]], ptr nonnull align 8 dereferenceable(16) [[D:%.*]]) #[[ATTR4]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 18 Lines
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[BN_ADDR]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[BN_ADDR]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[C_ADDR]], align 4 // CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[C_ADDR]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[VLA_ADDR2]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[VLA_ADDR2]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[VLA_ADDR4]], align 4 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[VLA_ADDR4]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = load ptr, ptr [[CN_ADDR]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load ptr, ptr [[CN_ADDR]], align 4
// CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[D_ADDR]], align 4 // CHECK2-NEXT: [[TMP7:%.*]] = load ptr, ptr [[D_ADDR]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP8:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP8]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP8]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[A_ADDR]], align 4 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], 1
// CHECK2-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4 // CHECK2-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x float], ptr [[TMP0]], i32 0, i32 2 // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x float], ptr [[TMP0]], i32 0, i32 2
// CHECK2-NEXT: [[TMP10:%.*]] = load float, ptr [[ARRAYIDX]], align 4 // CHECK2-NEXT: [[TMP10:%.*]] = load float, ptr [[ARRAYIDX]], align 4
Show All 27 Lines
// CHECK2-NEXT: [[CONV18:%.*]] = sext i8 [[TMP16]] to i32 // CHECK2-NEXT: [[CONV18:%.*]] = sext i8 [[TMP16]] to i32
// CHECK2-NEXT: [[ADD19:%.*]] = add nsw i32 [[CONV18]], 1 // CHECK2-NEXT: [[ADD19:%.*]] = add nsw i32 [[CONV18]], 1
// CHECK2-NEXT: [[CONV20:%.*]] = trunc i32 [[ADD19]] to i8 // CHECK2-NEXT: [[CONV20:%.*]] = trunc i32 [[ADD19]] to i8
// CHECK2-NEXT: store i8 [[CONV20]], ptr [[Y]], align 8 // CHECK2-NEXT: store i8 [[CONV20]], ptr [[Y]], align 8
// CHECK2-NEXT: [[CALL:%.*]] = call nonnull align 8 dereferenceable(8) ptr @_ZN2TTIxcEixEi(ptr nonnull align 8 dereferenceable(16) [[TMP7]], i32 0) #[[ATTR10:[0-9]+]] // CHECK2-NEXT: [[CALL:%.*]] = call nonnull align 8 dereferenceable(8) ptr @_ZN2TTIxcEixEi(ptr nonnull align 8 dereferenceable(16) [[TMP7]], i32 0) #[[ATTR10:[0-9]+]]
// CHECK2-NEXT: [[TMP17:%.*]] = load i64, ptr [[CALL]], align 8 // CHECK2-NEXT: [[TMP17:%.*]] = load i64, ptr [[CALL]], align 8
// CHECK2-NEXT: [[ADD21:%.*]] = add nsw i64 [[TMP17]], 1 // CHECK2-NEXT: [[ADD21:%.*]] = add nsw i64 [[TMP17]], 1
// CHECK2-NEXT: store i64 [[ADD21]], ptr [[CALL]], align 8 // CHECK2-NEXT: store i64 [[ADD21]], ptr [[CALL]], align 8
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@_ZN2TTIxcEixEi // CHECK2-LABEL: define {{[^@]+}}@_ZN2TTIxcEixEi
// CHECK2-SAME: (ptr nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32 [[I:%.*]]) #[[ATTR5:[0-9]+]] comdat align 2 { // CHECK2-SAME: (ptr nonnull align 8 dereferenceable(16) [[THIS:%.*]], i32 [[I:%.*]]) #[[ATTR5:[0-9]+]] comdat align 2 {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 13 Lines
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[AAA_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 // CHECK2-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: store i32 [[AAA]], ptr [[AAA_ADDR]], align 4 // CHECK2-NEXT: store i32 [[AAA]], ptr [[AAA_ADDR]], align 4
// CHECK2-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 4 // CHECK2-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[A_ADDR]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP2]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK2-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4 // CHECK2-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = load i16, ptr [[AA_ADDR]], align 2 // CHECK2-NEXT: [[TMP3:%.*]] = load i16, ptr [[AA_ADDR]], align 2
// CHECK2-NEXT: [[CONV2:%.*]] = sext i16 [[TMP3]] to i32 // CHECK2-NEXT: [[CONV:%.*]] = sext i16 [[TMP3]] to i32
// CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV2]], 1 // CHECK2-NEXT: [[ADD1:%.*]] = add nsw i32 [[CONV]], 1
// CHECK2-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i16 // CHECK2-NEXT: [[CONV2:%.*]] = trunc i32 [[ADD1]] to i16
// CHECK2-NEXT: store i16 [[CONV4]], ptr [[AA_ADDR]], align 2 // CHECK2-NEXT: store i16 [[CONV2]], ptr [[AA_ADDR]], align 2
// CHECK2-NEXT: [[TMP4:%.*]] = load i8, ptr [[AAA_ADDR]], align 1 // CHECK2-NEXT: [[TMP4:%.*]] = load i8, ptr [[AAA_ADDR]], align 1
// CHECK2-NEXT: [[CONV5:%.*]] = sext i8 [[TMP4]] to i32 // CHECK2-NEXT: [[CONV3:%.*]] = sext i8 [[TMP4]] to i32
// CHECK2-NEXT: [[ADD6:%.*]] = add nsw i32 [[CONV5]], 1 // CHECK2-NEXT: [[ADD4:%.*]] = add nsw i32 [[CONV3]], 1
// CHECK2-NEXT: [[CONV7:%.*]] = trunc i32 [[ADD6]] to i8 // CHECK2-NEXT: [[CONV5:%.*]] = trunc i32 [[ADD4]] to i8
// CHECK2-NEXT: store i8 [[CONV7]], ptr [[AAA_ADDR]], align 1 // CHECK2-NEXT: store i8 [[CONV5]], ptr [[AAA_ADDR]], align 1
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 2 // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 2
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
// CHECK2-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP5]], 1 // CHECK2-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP5]], 1
// CHECK2-NEXT: store i32 [[ADD8]], ptr [[ARRAYIDX]], align 4 // CHECK2-NEXT: store i32 [[ADD6]], ptr [[ARRAYIDX]], align 4
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l108 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2S12r1Ei_l108
// CHECK2-SAME: (ptr [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], ptr nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR4]] { // CHECK2-SAME: (ptr [[THIS:%.*]], i32 [[B:%.*]], i32 [[VLA:%.*]], i32 [[VLA1:%.*]], ptr nonnull align 2 dereferenceable(2) [[C:%.*]]) #[[ATTR4]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[VLA_ADDR2:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 // CHECK2-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK2-NEXT: store i32 [[B]], ptr [[B_ADDR]], align 4 // CHECK2-NEXT: store i32 [[B]], ptr [[B_ADDR]], align 4
// CHECK2-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 // CHECK2-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4
// CHECK2-NEXT: store i32 [[VLA1]], ptr [[VLA_ADDR2]], align 4 // CHECK2-NEXT: store i32 [[VLA1]], ptr [[VLA_ADDR2]], align 4
// CHECK2-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 4 // CHECK2-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[VLA_ADDR2]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[VLA_ADDR2]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[C_ADDR]], align 4 // CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[C_ADDR]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP4]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP4]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[B_ADDR]], align 4 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[B_ADDR]], align 4
// CHECK2-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP5]] to double // CHECK2-NEXT: [[CONV:%.*]] = sitofp i32 [[TMP5]] to double
// CHECK2-NEXT: [[ADD:%.*]] = fadd double [[CONV]], 1.500000e+00 // CHECK2-NEXT: [[ADD:%.*]] = fadd double [[CONV]], 1.500000e+00
// CHECK2-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S1:%.*]], ptr [[TMP0]], i32 0, i32 0 // CHECK2-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_S1:%.*]], ptr [[TMP0]], i32 0, i32 0
// CHECK2-NEXT: store double [[ADD]], ptr [[A]], align 8 // CHECK2-NEXT: store double [[ADD]], ptr [[A]], align 8
// CHECK2-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_S1]], ptr [[TMP0]], i32 0, i32 0 // CHECK2-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_S1]], ptr [[TMP0]], i32 0, i32 0
// CHECK2-NEXT: [[TMP6:%.*]] = load double, ptr [[A3]], align 8 // CHECK2-NEXT: [[TMP6:%.*]] = load double, ptr [[A3]], align 8
// CHECK2-NEXT: [[INC:%.*]] = fadd double [[TMP6]], 1.000000e+00 // CHECK2-NEXT: [[INC:%.*]] = fadd double [[TMP6]], 1.000000e+00
// CHECK2-NEXT: store double [[INC]], ptr [[A3]], align 8 // CHECK2-NEXT: store double [[INC]], ptr [[A3]], align 8
// CHECK2-NEXT: [[CONV4:%.*]] = fptosi double [[INC]] to i16 // CHECK2-NEXT: [[CONV4:%.*]] = fptosi double [[INC]] to i16
// CHECK2-NEXT: [[TMP7:%.*]] = mul nsw i32 1, [[TMP2]] // CHECK2-NEXT: [[TMP7:%.*]] = mul nsw i32 1, [[TMP2]]
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[TMP3]], i32 [[TMP7]] // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i16, ptr [[TMP3]], i32 [[TMP7]]
// CHECK2-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i16, ptr [[ARRAYIDX]], i32 1 // CHECK2-NEXT: [[ARRAYIDX5:%.*]] = getelementptr inbounds i16, ptr [[ARRAYIDX]], i32 1
// CHECK2-NEXT: store i16 [[CONV4]], ptr [[ARRAYIDX5]], align 2 // CHECK2-NEXT: store i16 [[CONV4]], ptr [[ARRAYIDX5]], align 2
// CHECK2-NEXT: [[A6:%.*]] = getelementptr inbounds [[STRUCT_S1]], ptr [[TMP0]], i32 0, i32 0 // CHECK2-NEXT: [[A6:%.*]] = getelementptr inbounds [[STRUCT_S1]], ptr [[TMP0]], i32 0, i32 0
// CHECK2-NEXT: [[TMP8:%.*]] = load double, ptr [[A6]], align 8 // CHECK2-NEXT: [[TMP8:%.*]] = load double, ptr [[A6]], align 8
// CHECK2-NEXT: [[CONV7:%.*]] = fptosi double [[TMP8]] to i32 // CHECK2-NEXT: [[CONV7:%.*]] = fptosi double [[TMP8]] to i32
// CHECK2-NEXT: [[A8:%.*]] = getelementptr inbounds [[STRUCT_S1]], ptr [[TMP0]], i32 0, i32 0 // CHECK2-NEXT: [[A8:%.*]] = getelementptr inbounds [[STRUCT_S1]], ptr [[TMP0]], i32 0, i32 0
// CHECK2-NEXT: [[CALL:%.*]] = call i32 @_Z3baziRd(i32 [[CONV7]], ptr nonnull align 8 dereferenceable(8) [[A8]]) #[[ATTR10]] // CHECK2-NEXT: [[CALL:%.*]] = call i32 @_Z3baziRd(i32 [[CONV7]], ptr nonnull align 8 dereferenceable(8) [[A8]]) #[[ATTR10]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@_Z3baziRd // CHECK2-LABEL: define {{[^@]+}}@_Z3baziRd
// CHECK2-SAME: (i32 [[F1:%.*]], ptr nonnull align 8 dereferenceable(8) [[A:%.*]]) #[[ATTR5]] { // CHECK2-SAME: (i32 [[F1:%.*]], ptr nonnull align 8 dereferenceable(8) [[A:%.*]]) #[[ATTR5]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [2 x ptr], align 4 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [2 x ptr], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB3:[0-9]+]]) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[F:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i32 4) // CHECK2-NEXT: [[F:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i32 4)
// CHECK2-NEXT: store i32 [[F1]], ptr [[F]], align 4 // CHECK2-NEXT: store i32 [[F1]], ptr [[F]], align 4
// CHECK2-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 // CHECK2-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[F]], ptr [[TMP2]], align 4 // CHECK2-NEXT: store ptr [[F]], ptr [[TMP2]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1 // CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK2-NEXT: store ptr [[TMP1]], ptr [[TMP4]], align 4 // CHECK2-NEXT: store ptr [[TMP1]], ptr [[TMP3]], align 4
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB3]], i32 [[TMP0]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 2) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP0]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 2)
// CHECK2-NEXT: [[TMP7:%.*]] = load i32, ptr [[F]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[F]], align 4
// CHECK2-NEXT: call void @__kmpc_free_shared(ptr [[F]], i32 4) // CHECK2-NEXT: call void @__kmpc_free_shared(ptr [[F]], i32 4)
// CHECK2-NEXT: ret i32 [[TMP7]] // CHECK2-NEXT: ret i32 [[TMP4]]
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16unreachable_callv_l142 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16unreachable_callv_l142
// CHECK2-SAME: () #[[ATTR4]] { // CHECK2-SAME: () #[[ATTR4]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: call void @_Z6asserti(i32 0) #[[ATTR11:[0-9]+]] // CHECK2-NEXT: call void @_Z6asserti(i32 0) #[[ATTR11:[0-9]+]]
// CHECK2-NEXT: unreachable // CHECK2-NEXT: unreachable
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: 1: // CHECK2: 1:
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l74 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l74
// CHECK2-SAME: (i32 [[A:%.*]], i32 [[AA:%.*]], ptr nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR4]] { // CHECK2-SAME: (i32 [[A:%.*]], i32 [[AA:%.*]], ptr nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR4]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 // CHECK2-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 4 // CHECK2-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[A_ADDR]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP2]], 1 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP2]], 1
// CHECK2-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4 // CHECK2-NEXT: store i32 [[ADD]], ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = load i16, ptr [[AA_ADDR]], align 2 // CHECK2-NEXT: [[TMP3:%.*]] = load i16, ptr [[AA_ADDR]], align 2
// CHECK2-NEXT: [[CONV1:%.*]] = sext i16 [[TMP3]] to i32 // CHECK2-NEXT: [[CONV:%.*]] = sext i16 [[TMP3]] to i32
// CHECK2-NEXT: [[ADD2:%.*]] = add nsw i32 [[CONV1]], 1 // CHECK2-NEXT: [[ADD1:%.*]] = add nsw i32 [[CONV]], 1
// CHECK2-NEXT: [[CONV3:%.*]] = trunc i32 [[ADD2]] to i16 // CHECK2-NEXT: [[CONV2:%.*]] = trunc i32 [[ADD1]] to i16
// CHECK2-NEXT: store i16 [[CONV3]], ptr [[AA_ADDR]], align 2 // CHECK2-NEXT: store i16 [[CONV2]], ptr [[AA_ADDR]], align 2
// CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 2 // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 2
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
// CHECK2-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP4]], 1 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP4]], 1
// CHECK2-NEXT: store i32 [[ADD4]], ptr [[ARRAYIDX]], align 4 // CHECK2-NEXT: store i32 [[ADD3]], ptr [[ARRAYIDX]], align 4
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK2-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr nonnull align 4 dereferenceable(4) [[F:%.*]], ptr nonnull align 8 dereferenceable(8) [[A:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]], ptr nonnull align 4 dereferenceable(4) [[F:%.*]], ptr nonnull align 8 dereferenceable(8) [[A:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 25 Lines
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2 // CHECK2-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2
// CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]]) // CHECK2-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]])
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i32 0 // CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i32 0
// CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP3]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i32 1 // CHECK2-NEXT: [[TMP5:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i32 1
// CHECK2-NEXT: [[TMP8:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 4
// CHECK2-NEXT: call void @__omp_outlined__1(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP5]], ptr [[TMP8]]) #[[ATTR2:[0-9]+]] // CHECK2-NEXT: call void @__omp_outlined__1(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP4]], ptr [[TMP6]]) #[[ATTR2:[0-9]+]]
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //

clang/test/OpenMP/nvptx_target_parallel_codegen.cpp

Show First 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
#endif #endif
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l30 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l30
// CHECK1-SAME: (ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK1-SAME: (ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP3]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP3]], align 8
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1)
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
Show All 20 Lines
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [3 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK1-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8 // CHECK1-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[B_ADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP3]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP3]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2]]) // CHECK1-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP5]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP5]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP7]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP7]], align 8
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP4]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP4]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3)
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
Show All 27 Lines
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l30 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l30
// CHECK2-SAME: (ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK2-SAME: (ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 4 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 4
// CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP3]], align 4 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP3]], align 4
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 1) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 1)
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 20 Lines
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [3 x ptr], align 4 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [3 x ptr], align 4
// CHECK2-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 // CHECK2-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 4 // CHECK2-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[B_ADDR]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[B_ADDR]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP3]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP3]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2]]) // CHECK2-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP5]], align 4 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP5]], align 4
// CHECK2-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK2-NEXT: store ptr [[TMP1]], ptr [[TMP7]], align 4 // CHECK2-NEXT: store ptr [[TMP1]], ptr [[TMP6]], align 4
// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2 // CHECK2-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
// CHECK2-NEXT: store ptr [[TMP2]], ptr [[TMP9]], align 4 // CHECK2-NEXT: store ptr [[TMP2]], ptr [[TMP7]], align 4
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP4]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP4]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3)
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 27 Lines

clang/test/OpenMP/nvptx_target_parallel_num_threads_codegen.cpp

Show First 20 Lines • Show All 47 Lines • ▼ Show 20 Lines
#endif #endif
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l25 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l25
// CHECK1-SAME: (ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK1-SAME: (ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP3]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP3]], align 8
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP2]], i32 1, i32 1024, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 1024, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1)
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
Show All 22 Lines
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [3 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK1-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8 // CHECK1-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8
// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[B_ADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP3]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP3]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2]]) // CHECK1-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP6]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP8]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP8]], align 8
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP4]], i32 1, i32 [[TMP5]], i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP4]], i32 1, i32 [[TMP5]], i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3)
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
Show All 27 Lines
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l25 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l25
// CHECK2-SAME: (ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK2-SAME: (ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 4 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 4
// CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP3]], align 4 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP3]], align 4
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP2]], i32 1, i32 1024, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 1) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 1024, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 1)
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 22 Lines
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [3 x ptr], align 4 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [3 x ptr], align 4
// CHECK2-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 // CHECK2-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 4 // CHECK2-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 4
// CHECK2-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK2-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[B_ADDR]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[B_ADDR]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP3]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP3]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2]]) // CHECK2-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP6]], align 4 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP6]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1 // CHECK2-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK2-NEXT: store ptr [[TMP1]], ptr [[TMP8]], align 4 // CHECK2-NEXT: store ptr [[TMP1]], ptr [[TMP7]], align 4
// CHECK2-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2 // CHECK2-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
// CHECK2-NEXT: store ptr [[TMP2]], ptr [[TMP10]], align 4 // CHECK2-NEXT: store ptr [[TMP2]], ptr [[TMP8]], align 4
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP4]], i32 1, i32 [[TMP5]], i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP4]], i32 1, i32 [[TMP5]], i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3)
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 27 Lines

clang/test/OpenMP/nvptx_target_parallel_reduction_codegen.cpp

Show First 20 Lines • Show All 48 Lines • ▼ Show 20 Linesint bar(int n){
a += ftemplate<char>(n); a += ftemplate<char>(n);
return a; return a;
} }
// CHECK: define {{.*}}void {{@__omp_offloading_.+template.+l27}}( // CHECK: define {{.*}}void {{@__omp_offloading_.+template.+l27}}(
// //
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2, i1 true) // CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2)
// //
// //
// define internal void [[PFN]]( // define internal void [[PFN]](
// CHECK: store double {{[0\.e\+]+}}, double* [[E:%.+]], align // CHECK: store double {{[0\.e\+]+}}, double* [[E:%.+]], align
// CHECK: [[EV:%.+]] = load double, double* [[E]], align // CHECK: [[EV:%.+]] = load double, double* [[E]], align
// CHECK: [[ADD:%.+]] = fadd double [[EV]], 5 // CHECK: [[ADD:%.+]] = fadd double [[EV]], 5
// CHECK: store double [[ADD]], double* [[E]], align // CHECK: store double [[ADD]], double* [[E]], align
// CHECK: [[PTR1:%.+]] = getelementptr inbounds [[RLT:.+]], [1 x i8*]* [[RL:%.+]], i{{32|64}} 0, i{{32|64}} 0 // CHECK: [[PTR1:%.+]] = getelementptr inbounds [[RLT:.+]], [1 x i8*]* [[RL:%.+]], i{{32|64}} 0, i{{32|64}} 0
▲ Show 20 Lines • Show All 161 Lines • ▼ Show 20 Lines
// CHECK: [[READ_CONT]] // CHECK: [[READ_CONT]]
// CHECK: [[NEXT:%.+]] = add nsw i32 [[CNT]], 1 // CHECK: [[NEXT:%.+]] = add nsw i32 [[CNT]], 1
// CHECK: store i32 [[NEXT]], i32* [[CNT_ADDR]], // CHECK: store i32 [[NEXT]], i32* [[CNT_ADDR]],
// CHECK: br label // CHECK: br label
// CHECK: ret // CHECK: ret
// CHECK: define {{.*}}void {{@__omp_offloading_.+template.+l32}}( // CHECK: define {{.*}}void {{@__omp_offloading_.+template.+l32}}(
// //
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2, i1 true) // CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2)
// //
// //
// define internal void [[PFN1]]( // define internal void [[PFN1]](
// CHECK: store float {{1\.[0e\+]+}}, float* [[D:%.+]], align // CHECK: store float {{1\.[0e\+]+}}, float* [[D:%.+]], align
// CHECK: [[C_VAL:%.+]] = load i8, i8* [[C:%.+]], align // CHECK: [[C_VAL:%.+]] = load i8, i8* [[C:%.+]], align
// CHECK: [[CONV:%.+]] = sext i8 [[C_VAL]] to i32 // CHECK: [[CONV:%.+]] = sext i8 [[C_VAL]] to i32
// CHECK: [[XOR:%.+]] = xor i32 [[CONV]], 2 // CHECK: [[XOR:%.+]] = xor i32 [[CONV]], 2
// CHECK: [[TRUNC:%.+]] = trunc i32 [[XOR]] to i8 // CHECK: [[TRUNC:%.+]] = trunc i32 [[XOR]] to i8
▲ Show 20 Lines • Show All 239 Lines • ▼ Show 20 Lines
// CHECK: [[READ_ELSE]] // CHECK: [[READ_ELSE]]
// CHECK: br label {{%?}}[[READ_CONT]] // CHECK: br label {{%?}}[[READ_CONT]]
// //
// CHECK: [[READ_CONT]] // CHECK: [[READ_CONT]]
// CHECK: ret // CHECK: ret
// CHECK: define {{.*}}void {{@__omp_offloading_.+template.+l38}}( // CHECK: define {{.*}}void {{@__omp_offloading_.+template.+l38}}(
// //
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2, i1 true) // CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2)
// //
// //
// define internal void [[PFN2]]( // define internal void [[PFN2]](
// CHECK: store i32 0, i32* [[A:%.+]], align // CHECK: store i32 0, i32* [[A:%.+]], align
// CHECK: store i16 -32768, i16* [[B:%.+]], align // CHECK: store i16 -32768, i16* [[B:%.+]], align
// CHECK: [[A_VAL:%.+]] = load i32, i32* [[A:%.+]], align // CHECK: [[A_VAL:%.+]] = load i32, i32* [[A:%.+]], align
// CHECK: [[OR:%.+]] = or i32 [[A_VAL]], 1 // CHECK: [[OR:%.+]] = or i32 [[A_VAL]], 1
// CHECK: store i32 [[OR]], i32* [[A]], align // CHECK: store i32 [[OR]], i32* [[A]], align
▲ Show 20 Lines • Show All 290 Lines • Show Last 20 Lines

clang/test/OpenMP/nvptx_target_parallel_reduction_codegen_tbaa_PR46146.cpp

Show All 29 Linesvoid test() {
complex_reduction<double>(); complex_reduction<double>();
} }
#endif #endif
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIfEvv_l16 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIfEvv_l16
// CHECK1-SAME: () #[[ATTR0:[0-9]+]] { // CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA8:![0-9]+]] // CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA8:![0-9]+]]
// CHECK1-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]]) #[[ATTR6:[0-9]+]] // CHECK1-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]]) #[[ATTR6:[0-9]+]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK1-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 467 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIdEvv_l16 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z17complex_reductionIdEvv_l16
// CHECK1-SAME: () #[[ATTR0]] { // CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA8]] // CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA8]]
// CHECK1-NEXT: call void @__omp_outlined__2(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]]) #[[ATTR6]] // CHECK1-NEXT: call void @__omp_outlined__2(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]]) #[[ATTR6]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__2 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__2
// CHECK1-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias [[DOTGLOBAL_TID_:%.*]], ptr noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 568 Lines • Show Last 20 Lines

clang/test/OpenMP/nvptx_target_printf_codegen.c

Show First 20 Lines • Show All 42 Lines • ▼ Show 20 Lines
// //
// //
// //
// CHECK-64-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_CheckSimple_l13 // CHECK-64-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_CheckSimple_l13
// CHECK-64-SAME: () #[[ATTR0:[0-9]+]] { // CHECK-64-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK-64-NEXT: entry: // CHECK-64-NEXT: entry:
// CHECK-64-NEXT: [[FMT:%.*]] = alloca ptr, align 8 // CHECK-64-NEXT: [[FMT:%.*]] = alloca ptr, align 8
// CHECK-64-NEXT: [[TMP:%.*]] = alloca [[PRINTF_ARGS:%.*]], align 8 // CHECK-64-NEXT: [[TMP:%.*]] = alloca [[PRINTF_ARGS:%.*]], align 8
// CHECK-64-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK-64-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK-64-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK-64-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK-64-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK-64-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK-64: user_code.entry: // CHECK-64: user_code.entry:
// CHECK-64-NEXT: store ptr @.str, ptr [[FMT]], align 8 // CHECK-64-NEXT: store ptr @.str, ptr [[FMT]], align 8
// CHECK-64-NEXT: [[TMP1:%.*]] = load ptr, ptr [[FMT]], align 8 // CHECK-64-NEXT: [[TMP1:%.*]] = load ptr, ptr [[FMT]], align 8
// CHECK-64-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[PRINTF_ARGS]], ptr [[TMP]], i32 0, i32 0 // CHECK-64-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[PRINTF_ARGS]], ptr [[TMP]], i32 0, i32 0
// CHECK-64-NEXT: store i32 1, ptr [[TMP2]], align 4 // CHECK-64-NEXT: store i32 1, ptr [[TMP2]], align 4
// CHECK-64-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[PRINTF_ARGS]], ptr [[TMP]], i32 0, i32 1 // CHECK-64-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[PRINTF_ARGS]], ptr [[TMP]], i32 0, i32 1
// CHECK-64-NEXT: store i64 2, ptr [[TMP3]], align 8 // CHECK-64-NEXT: store i64 2, ptr [[TMP3]], align 8
// CHECK-64-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[PRINTF_ARGS]], ptr [[TMP]], i32 0, i32 2 // CHECK-64-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[PRINTF_ARGS]], ptr [[TMP]], i32 0, i32 2
// CHECK-64-NEXT: store double 3.000000e+00, ptr [[TMP4]], align 8 // CHECK-64-NEXT: store double 3.000000e+00, ptr [[TMP4]], align 8
// CHECK-64-NEXT: [[TMP6:%.*]] = call i32 @__llvm_omp_vprintf(ptr [[TMP1]], ptr [[TMP]], i32 24) // CHECK-64-NEXT: [[TMP6:%.*]] = call i32 @__llvm_omp_vprintf(ptr [[TMP1]], ptr [[TMP]], i32 24)
// CHECK-64-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK-64-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK-64-NEXT: ret void // CHECK-64-NEXT: ret void
// CHECK-64: worker.exit: // CHECK-64: worker.exit:
// CHECK-64-NEXT: ret void // CHECK-64-NEXT: ret void
// //
// //
// CHECK-64-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_CheckNoArgs_l25 // CHECK-64-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_CheckNoArgs_l25
// CHECK-64-SAME: () #[[ATTR0]] { // CHECK-64-SAME: () #[[ATTR0]] {
// CHECK-64-NEXT: entry: // CHECK-64-NEXT: entry:
// CHECK-64-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK-64-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK-64-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK-64-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK-64-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK-64-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK-64: user_code.entry: // CHECK-64: user_code.entry:
// CHECK-64-NEXT: [[TMP1:%.*]] = call i32 @__llvm_omp_vprintf(ptr @.str1, ptr null, i32 0) // CHECK-64-NEXT: [[TMP1:%.*]] = call i32 @__llvm_omp_vprintf(ptr @.str1, ptr null, i32 0)
// CHECK-64-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK-64-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK-64-NEXT: ret void // CHECK-64-NEXT: ret void
// CHECK-64: worker.exit: // CHECK-64: worker.exit:
// CHECK-64-NEXT: ret void // CHECK-64-NEXT: ret void
// //
// //
// CHECK-64-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_CheckAllocaIsInEntryBlock_l36 // CHECK-64-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_CheckAllocaIsInEntryBlock_l36
// CHECK-64-SAME: (i64 noundef [[FOO:%.*]]) #[[ATTR0]] { // CHECK-64-SAME: (i64 noundef [[FOO:%.*]]) #[[ATTR0]] {
// CHECK-64-NEXT: entry: // CHECK-64-NEXT: entry:
// CHECK-64-NEXT: [[FOO_ADDR:%.*]] = alloca i64, align 8 // CHECK-64-NEXT: [[FOO_ADDR:%.*]] = alloca i64, align 8
// CHECK-64-NEXT: [[TMP:%.*]] = alloca [[PRINTF_ARGS_0:%.*]], align 8 // CHECK-64-NEXT: [[TMP:%.*]] = alloca [[PRINTF_ARGS_0:%.*]], align 8
// CHECK-64-NEXT: store i64 [[FOO]], ptr [[FOO_ADDR]], align 8 // CHECK-64-NEXT: store i64 [[FOO]], ptr [[FOO_ADDR]], align 8
// CHECK-64-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK-64-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK-64-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK-64-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK-64-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK-64-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK-64: user_code.entry: // CHECK-64: user_code.entry:
// CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[FOO_ADDR]], align 4 // CHECK-64-NEXT: [[TMP1:%.*]] = load i32, ptr [[FOO_ADDR]], align 4
// CHECK-64-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP1]], 0 // CHECK-64-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP1]], 0
// CHECK-64-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[IF_END:%.*]] // CHECK-64-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[IF_END:%.*]]
// CHECK-64: if.then: // CHECK-64: if.then:
// CHECK-64-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[PRINTF_ARGS_0]], ptr [[TMP]], i32 0, i32 0 // CHECK-64-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[PRINTF_ARGS_0]], ptr [[TMP]], i32 0, i32 0
// CHECK-64-NEXT: store i32 42, ptr [[TMP2]], align 4 // CHECK-64-NEXT: store i32 42, ptr [[TMP2]], align 4
// CHECK-64-NEXT: [[TMP4:%.*]] = call i32 @__llvm_omp_vprintf(ptr @.str2, ptr [[TMP]], i32 4) // CHECK-64-NEXT: [[TMP4:%.*]] = call i32 @__llvm_omp_vprintf(ptr @.str2, ptr [[TMP]], i32 4)
// CHECK-64-NEXT: br label [[IF_END]] // CHECK-64-NEXT: br label [[IF_END]]
// CHECK-64: worker.exit: // CHECK-64: worker.exit:
// CHECK-64-NEXT: ret void // CHECK-64-NEXT: ret void
// CHECK-64: if.end: // CHECK-64: if.end:
// CHECK-64-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK-64-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK-64-NEXT: ret void // CHECK-64-NEXT: ret void
// //
// //
// //
// //
// //
// CHECK-32-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_CheckSimple_l13 // CHECK-32-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_CheckSimple_l13
// CHECK-32-SAME: () #[[ATTR0:[0-9]+]] { // CHECK-32-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK-32-NEXT: entry: // CHECK-32-NEXT: entry:
// CHECK-32-NEXT: [[FMT:%.*]] = alloca ptr, align 4 // CHECK-32-NEXT: [[FMT:%.*]] = alloca ptr, align 4
// CHECK-32-NEXT: [[TMP:%.*]] = alloca [[PRINTF_ARGS:%.*]], align 8 // CHECK-32-NEXT: [[TMP:%.*]] = alloca [[PRINTF_ARGS:%.*]], align 8
// CHECK-32-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK-32-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK-32-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK-32-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK-32-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK-32-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK-32: user_code.entry: // CHECK-32: user_code.entry:
// CHECK-32-NEXT: store ptr @.str, ptr [[FMT]], align 4 // CHECK-32-NEXT: store ptr @.str, ptr [[FMT]], align 4
// CHECK-32-NEXT: [[TMP1:%.*]] = load ptr, ptr [[FMT]], align 4 // CHECK-32-NEXT: [[TMP1:%.*]] = load ptr, ptr [[FMT]], align 4
// CHECK-32-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[PRINTF_ARGS]], ptr [[TMP]], i32 0, i32 0 // CHECK-32-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[PRINTF_ARGS]], ptr [[TMP]], i32 0, i32 0
// CHECK-32-NEXT: store i32 1, ptr [[TMP2]], align 4 // CHECK-32-NEXT: store i32 1, ptr [[TMP2]], align 4
// CHECK-32-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[PRINTF_ARGS]], ptr [[TMP]], i32 0, i32 1 // CHECK-32-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[PRINTF_ARGS]], ptr [[TMP]], i32 0, i32 1
// CHECK-32-NEXT: store i64 2, ptr [[TMP3]], align 8 // CHECK-32-NEXT: store i64 2, ptr [[TMP3]], align 8
// CHECK-32-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[PRINTF_ARGS]], ptr [[TMP]], i32 0, i32 2 // CHECK-32-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[PRINTF_ARGS]], ptr [[TMP]], i32 0, i32 2
// CHECK-32-NEXT: store double 3.000000e+00, ptr [[TMP4]], align 8 // CHECK-32-NEXT: store double 3.000000e+00, ptr [[TMP4]], align 8
// CHECK-32-NEXT: [[TMP6:%.*]] = call i32 @__llvm_omp_vprintf(ptr [[TMP1]], ptr [[TMP]], i32 24) // CHECK-32-NEXT: [[TMP6:%.*]] = call i32 @__llvm_omp_vprintf(ptr [[TMP1]], ptr [[TMP]], i32 24)
// CHECK-32-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK-32-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK-32-NEXT: ret void // CHECK-32-NEXT: ret void
// CHECK-32: worker.exit: // CHECK-32: worker.exit:
// CHECK-32-NEXT: ret void // CHECK-32-NEXT: ret void
// //
// //
// CHECK-32-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_CheckNoArgs_l25 // CHECK-32-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_CheckNoArgs_l25
// CHECK-32-SAME: () #[[ATTR0]] { // CHECK-32-SAME: () #[[ATTR0]] {
// CHECK-32-NEXT: entry: // CHECK-32-NEXT: entry:
// CHECK-32-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK-32-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK-32-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK-32-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK-32-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK-32-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK-32: user_code.entry: // CHECK-32: user_code.entry:
// CHECK-32-NEXT: [[TMP1:%.*]] = call i32 @__llvm_omp_vprintf(ptr @.str1, ptr null, i32 0) // CHECK-32-NEXT: [[TMP1:%.*]] = call i32 @__llvm_omp_vprintf(ptr @.str1, ptr null, i32 0)
// CHECK-32-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK-32-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK-32-NEXT: ret void // CHECK-32-NEXT: ret void
// CHECK-32: worker.exit: // CHECK-32: worker.exit:
// CHECK-32-NEXT: ret void // CHECK-32-NEXT: ret void
// //
// //
// CHECK-32-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_CheckAllocaIsInEntryBlock_l36 // CHECK-32-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_CheckAllocaIsInEntryBlock_l36
// CHECK-32-SAME: (i32 noundef [[FOO:%.*]]) #[[ATTR0]] { // CHECK-32-SAME: (i32 noundef [[FOO:%.*]]) #[[ATTR0]] {
// CHECK-32-NEXT: entry: // CHECK-32-NEXT: entry:
// CHECK-32-NEXT: [[FOO_ADDR:%.*]] = alloca i32, align 4 // CHECK-32-NEXT: [[FOO_ADDR:%.*]] = alloca i32, align 4
// CHECK-32-NEXT: [[TMP:%.*]] = alloca [[PRINTF_ARGS_0:%.*]], align 8 // CHECK-32-NEXT: [[TMP:%.*]] = alloca [[PRINTF_ARGS_0:%.*]], align 8
// CHECK-32-NEXT: store i32 [[FOO]], ptr [[FOO_ADDR]], align 4 // CHECK-32-NEXT: store i32 [[FOO]], ptr [[FOO_ADDR]], align 4
// CHECK-32-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK-32-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK-32-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK-32-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK-32-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK-32-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK-32: user_code.entry: // CHECK-32: user_code.entry:
// CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[FOO_ADDR]], align 4 // CHECK-32-NEXT: [[TMP1:%.*]] = load i32, ptr [[FOO_ADDR]], align 4
// CHECK-32-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP1]], 0 // CHECK-32-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP1]], 0
// CHECK-32-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[IF_END:%.*]] // CHECK-32-NEXT: br i1 [[TOBOOL]], label [[IF_THEN:%.*]], label [[IF_END:%.*]]
// CHECK-32: if.then: // CHECK-32: if.then:
// CHECK-32-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[PRINTF_ARGS_0]], ptr [[TMP]], i32 0, i32 0 // CHECK-32-NEXT: [[TMP2:%.*]] = getelementptr inbounds [[PRINTF_ARGS_0]], ptr [[TMP]], i32 0, i32 0
// CHECK-32-NEXT: store i32 42, ptr [[TMP2]], align 4 // CHECK-32-NEXT: store i32 42, ptr [[TMP2]], align 4
// CHECK-32-NEXT: [[TMP4:%.*]] = call i32 @__llvm_omp_vprintf(ptr @.str2, ptr [[TMP]], i32 4) // CHECK-32-NEXT: [[TMP4:%.*]] = call i32 @__llvm_omp_vprintf(ptr @.str2, ptr [[TMP]], i32 4)
// CHECK-32-NEXT: br label [[IF_END]] // CHECK-32-NEXT: br label [[IF_END]]
// CHECK-32: worker.exit: // CHECK-32: worker.exit:
// CHECK-32-NEXT: ret void // CHECK-32-NEXT: ret void
// CHECK-32: if.end: // CHECK-32: if.end:
// CHECK-32-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK-32-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK-32-NEXT: ret void // CHECK-32-NEXT: ret void
// //

clang/test/OpenMP/nvptx_target_simd_codegen.cpp

Show First 20 Lines • Show All 55 Lines • ▼ Show 20 Linesint bar(int n){
int a = 0; int a = 0;
a += ftemplate<int>(n); a += ftemplate<int>(n);
return a; return a;
} }
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+l32}}( // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+l32}}(
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-NOT: call void @__kmpc_for_static_init // CHECK-NOT: call void @__kmpc_for_static_init
// CHECK-NOT: call void @__kmpc_for_static_fini // CHECK-NOT: call void @__kmpc_for_static_fini
// CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2, i1 true) // CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2)
// CHECK: ret void // CHECK: ret void
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+l37}}( // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+l37}}(
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-NOT: call void @__kmpc_for_static_init // CHECK-NOT: call void @__kmpc_for_static_init
// CHECK-NOT: call void @__kmpc_for_static_fini // CHECK-NOT: call void @__kmpc_for_static_fini
// CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2, i1 true) // CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2)
// CHECK: ret void // CHECK: ret void
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+l42}}( // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+l42}}(
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-NOT: call void @__kmpc_for_static_init // CHECK-NOT: call void @__kmpc_for_static_init
// CHECK-NOT: call void @__kmpc_for_static_fini // CHECK-NOT: call void @__kmpc_for_static_fini
// CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2, i1 true) // CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2)
// CHECK: ret void // CHECK: ret void
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+l47}}( // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+l47}}(
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK-NOT: call void @__kmpc_for_static_init // CHECK-NOT: call void @__kmpc_for_static_init
// CHECK-NOT: call void @__kmpc_for_static_fini // CHECK-NOT: call void @__kmpc_for_static_fini
// CHECK-NOT: call void @__kmpc_nvptx_end_reduce_nowait( // CHECK-NOT: call void @__kmpc_nvptx_end_reduce_nowait(
// CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2, i1 true) // CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2)
// CHECK: ret void // CHECK: ret void
#endif #endif

clang/test/OpenMP/nvptx_target_teams_codegen.cpp

Show First 20 Lines • Show All 51 Lines • ▼ Show 20 Lines
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l23 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l23
// CHECK1-SAME: (i64 noundef [[A:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK1-SAME: (i64 noundef [[A:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[A_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[A_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8 // CHECK1-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP2:%.*]] = load i8, ptr [[A_ADDR]], align 1 // CHECK1-NEXT: [[TMP2:%.*]] = load i8, ptr [[A_ADDR]], align 1
// CHECK1-NEXT: store i8 [[TMP2]], ptr [[A_CASTED]], align 1 // CHECK1-NEXT: store i8 [[TMP2]], ptr [[A_CASTED]], align 1
// CHECK1-NEXT: [[TMP3:%.*]] = load i64, ptr [[A_CASTED]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load i64, ptr [[A_CASTED]], align 8
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK1-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP3]]) #[[ATTR2:[0-9]+]] // CHECK1-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP3]]) #[[ATTR2:[0-9]+]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[A:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[A:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
Show All 10 Lines
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l28 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l28
// CHECK1-SAME: (i64 noundef [[AA:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (i64 noundef [[AA:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AA_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP2:%.*]] = load i16, ptr [[AA_ADDR]], align 2 // CHECK1-NEXT: [[TMP2:%.*]] = load i16, ptr [[AA_ADDR]], align 2
// CHECK1-NEXT: store i16 [[TMP2]], ptr [[AA_CASTED]], align 2 // CHECK1-NEXT: store i16 [[TMP2]], ptr [[AA_CASTED]], align 2
// CHECK1-NEXT: [[TMP3:%.*]] = load i64, ptr [[AA_CASTED]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load i64, ptr [[AA_CASTED]], align 8
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK1-NEXT: call void @__omp_outlined__1(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP3]]) #[[ATTR2]] // CHECK1-NEXT: call void @__omp_outlined__1(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP3]]) #[[ATTR2]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[AA:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[AA:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
Show All 10 Lines
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33
// CHECK1-SAME: (i64 noundef [[AA:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (i64 noundef [[AA:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AA_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]]) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP2:%.*]] = load i16, ptr [[AA_ADDR]], align 2 // CHECK1-NEXT: [[TMP2:%.*]] = load i16, ptr [[AA_ADDR]], align 2
// CHECK1-NEXT: store i16 [[TMP2]], ptr [[AA_CASTED]], align 2 // CHECK1-NEXT: store i16 [[TMP2]], ptr [[AA_CASTED]], align 2
// CHECK1-NEXT: [[TMP3:%.*]] = load i64, ptr [[AA_CASTED]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load i64, ptr [[AA_CASTED]], align 8
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK1-NEXT: call void @__omp_outlined__2(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP3]]) #[[ATTR2]] // CHECK1-NEXT: call void @__omp_outlined__2(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP3]]) #[[ATTR2]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__2 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__2
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[AA:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[AA:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: store i64 [[AA]], ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[AA_ADDR]], ptr [[TMP0]], align 8 // CHECK1-NEXT: store ptr [[AA_ADDR]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP3]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__3, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__3, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__3 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__3
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP1]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP4]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__4, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP3]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__4, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__4 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__4
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
Show All 9 Lines
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l23 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l23
// CHECK2-SAME: (i32 noundef [[A:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK2-SAME: (i32 noundef [[A:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[A_CASTED:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[A_CASTED:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 // CHECK2-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP2:%.*]] = load i8, ptr [[A_ADDR]], align 1 // CHECK2-NEXT: [[TMP2:%.*]] = load i8, ptr [[A_ADDR]], align 1
// CHECK2-NEXT: store i8 [[TMP2]], ptr [[A_CASTED]], align 1 // CHECK2-NEXT: store i8 [[TMP2]], ptr [[A_CASTED]], align 1
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[A_CASTED]], align 4 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[A_CASTED]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK2-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP3]]) #[[ATTR2:[0-9]+]] // CHECK2-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP3]]) #[[ATTR2:[0-9]+]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[A:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[A:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 10 Lines
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l28 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l28
// CHECK2-SAME: (i32 noundef [[AA:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (i32 noundef [[AA:%.*]]) #[[ATTR0]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[AA_CASTED:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[AA_CASTED:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP2:%.*]] = load i16, ptr [[AA_ADDR]], align 2 // CHECK2-NEXT: [[TMP2:%.*]] = load i16, ptr [[AA_ADDR]], align 2
// CHECK2-NEXT: store i16 [[TMP2]], ptr [[AA_CASTED]], align 2 // CHECK2-NEXT: store i16 [[TMP2]], ptr [[AA_CASTED]], align 2
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[AA_CASTED]], align 4 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[AA_CASTED]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK2-NEXT: call void @__omp_outlined__1(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP3]]) #[[ATTR2]] // CHECK2-NEXT: call void @__omp_outlined__1(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP3]]) #[[ATTR2]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[AA:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[AA:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 10 Lines
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l33
// CHECK2-SAME: (i32 noundef [[AA:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (i32 noundef [[AA:%.*]]) #[[ATTR0]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[AA_CASTED:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[AA_CASTED:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]]) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP2:%.*]] = load i16, ptr [[AA_ADDR]], align 2 // CHECK2-NEXT: [[TMP2:%.*]] = load i16, ptr [[AA_ADDR]], align 2
// CHECK2-NEXT: store i16 [[TMP2]], ptr [[AA_CASTED]], align 2 // CHECK2-NEXT: store i16 [[TMP2]], ptr [[AA_CASTED]], align 2
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[AA_CASTED]], align 4 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[AA_CASTED]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK2-NEXT: call void @__omp_outlined__2(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP3]]) #[[ATTR2]] // CHECK2-NEXT: call void @__omp_outlined__2(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP3]]) #[[ATTR2]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__2 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__2
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[AA:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[AA:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 4 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 4
// CHECK2-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK2-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 // CHECK2-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: store i32 [[AA]], ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[AA_ADDR]], ptr [[TMP0]], align 4 // CHECK2-NEXT: store ptr [[AA_ADDR]], ptr [[TMP0]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP3]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__3, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 1) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__3, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__3 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__3
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 4 // CHECK2-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 4
// CHECK2-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK2-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 // CHECK2-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP1]], align 4 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP1]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP4]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__4, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 1) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP3]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__4, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__4 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__4
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 2 dereferenceable(2) [[AA:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK2-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 // CHECK2-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: store i16 1, ptr [[TMP0]], align 2 // CHECK2-NEXT: store i16 1, ptr [[TMP0]], align 2
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //

clang/test/OpenMP/nvptx_target_teams_distribute_codegen.cpp

Show All 31 Lines
} }
#endif #endif
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l16 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l16
// CHECK1-SAME: () #[[ATTR0:[0-9]+]] { // CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK1-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]]) #[[ATTR5:[0-9]+]] // CHECK1-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]]) #[[ATTR5:[0-9]+]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 42 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[I]], ptr [[TMP8]], align 8 // CHECK1-NEXT: store ptr [[I]], ptr [[TMP8]], align 8
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 1) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i64 1)
// 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: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1 // CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP9]], 1
// CHECK1-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]]) // CHECK1-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK1-NEXT: call void @__kmpc_free_shared(ptr [[I]], i64 4) // CHECK1-NEXT: call void @__kmpc_free_shared(ptr [[I]], i64 4)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
Show All 23 Lines
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2 // CHECK1-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2
// CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK1-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]]) // CHECK1-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]])
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP3]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 8
// CHECK1-NEXT: call void @__omp_outlined__1(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP5]]) #[[ATTR5]] // CHECK1-NEXT: call void @__omp_outlined__1(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP4]]) #[[ATTR5]]
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l16 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIcET_i_l16
// CHECK2-SAME: () #[[ATTR0:[0-9]+]] { // CHECK2-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK2-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]]) #[[ATTR5:[0-9]+]] // CHECK2-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]]) #[[ATTR5:[0-9]+]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
▲ Show 20 Lines • Show All 42 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: store i32 [[ADD]], ptr [[I]], align 4 // CHECK2-NEXT: store i32 [[ADD]], ptr [[I]], align 4
// CHECK2-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP8:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK2-NEXT: store ptr [[I]], ptr [[TMP8]], align 4 // CHECK2-NEXT: store ptr [[I]], ptr [[TMP8]], align 4
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 1) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr @__omp_outlined__1_wrapper, ptr [[CAPTURED_VARS_ADDRS]], i32 1)
// 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: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1 // CHECK2-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP9]], 1
// CHECK2-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 // CHECK2-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]]) // CHECK2-NEXT: call void @__kmpc_distribute_static_fini(ptr @[[GLOB2]], i32 [[TMP1]])
// CHECK2-NEXT: call void @__kmpc_free_shared(ptr [[I]], i32 4) // CHECK2-NEXT: call void @__kmpc_free_shared(ptr [[I]], i32 4)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
Show All 23 Lines
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[GLOBAL_ARGS:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2 // CHECK2-NEXT: store i16 [[TMP0]], ptr [[DOTADDR]], align 2
// CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK2-NEXT: store i32 [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]]) // CHECK2-NEXT: call void @__kmpc_get_shared_variables(ptr [[GLOBAL_ARGS]])
// CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 4 // CHECK2-NEXT: [[TMP2:%.*]] = load ptr, ptr [[GLOBAL_ARGS]], align 4
// CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i32 0 // CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds ptr, ptr [[TMP2]], i32 0
// CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP3]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load ptr, ptr [[TMP3]], align 4
// CHECK2-NEXT: call void @__omp_outlined__1(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP5]]) #[[ATTR5]] // CHECK2-NEXT: call void @__omp_outlined__1(ptr [[DOTADDR1]], ptr [[DOTZERO_ADDR]], ptr [[TMP4]]) #[[ATTR5]]
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //

clang/test/OpenMP/nvptx_target_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 79 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[L_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[L_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK1-NEXT: store i64 [[L]], ptr [[L_ADDR]], align 8 // CHECK1-NEXT: store i64 [[L]], ptr [[L_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4:[0-9]+]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 // CHECK1-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[L_ADDR]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[L_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], ptr [[L_CASTED]], align 4 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[L_CASTED]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[L_CASTED]], align 8 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[L_CASTED]], align 8
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK1-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]], i64 [[TMP6]]) #[[ATTR3:[0-9]+]] // CHECK1-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]], i64 [[TMP6]]) #[[ATTR3:[0-9]+]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[A:%.*]], i64 noundef [[L:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[A:%.*]], i64 noundef [[L:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 86 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store ptr [[TMP27]], ptr [[TMP26]], align 8 // CHECK1-NEXT: store ptr [[TMP27]], ptr [[TMP26]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP28]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP28]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
// CHECK1-NEXT: [[TMP30:%.*]] = inttoptr i64 [[TMP21]] to ptr // CHECK1-NEXT: [[TMP30:%.*]] = inttoptr i64 [[TMP21]] to ptr
// CHECK1-NEXT: store ptr [[TMP30]], ptr [[TMP29]], align 8 // CHECK1-NEXT: store ptr [[TMP30]], ptr [[TMP29]], align 8
// CHECK1-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4 // CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP32]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP32]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5)
// 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: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP33]], [[TMP34]] // CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP33]], [[TMP34]]
// CHECK1-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 177 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK1-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 // CHECK1-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK1-NEXT: call void @__omp_outlined__2(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]]) #[[ATTR3]] // CHECK1-NEXT: call void @__omp_outlined__2(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]]) #[[ATTR3]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__2 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__2
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 2 dereferenceable(2000) [[AA:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 2 dereferenceable(2000) [[AA:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 77 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 8 // CHECK1-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK1-NEXT: [[TMP25:%.*]] = inttoptr i64 [[TMP19]] to ptr // CHECK1-NEXT: [[TMP25:%.*]] = inttoptr i64 [[TMP19]] to ptr
// CHECK1-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 8 // CHECK1-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4 // CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__3, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__3, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 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: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP29]], [[TMP30]] // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP29]], [[TMP30]]
// CHECK1-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 127 Lines • ▼ Show 20 Lines
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l39 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l39
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8 // CHECK1-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK1-NEXT: call void @__omp_outlined__4(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]]) #[[ATTR3]] // CHECK1-NEXT: call void @__omp_outlined__4(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]]) #[[ATTR3]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__4 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__4
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 46 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK1-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP8]] to ptr // CHECK1-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP8]] to ptr
// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8 // CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK1-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP10]] to ptr // CHECK1-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP10]] to ptr
// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8 // CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 8
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__5, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__5, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3)
// 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: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 100 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[F_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[F_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[F_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[F_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 // CHECK1-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8
// CHECK1-NEXT: store i64 [[F]], ptr [[F_ADDR]], align 8 // CHECK1-NEXT: store i64 [[F]], ptr [[F_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[F_ADDR]], align 4 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[F_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP3]], ptr [[F_CASTED]], align 4 // CHECK1-NEXT: store i32 [[TMP3]], ptr [[F_CASTED]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[F_CASTED]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[F_CASTED]], align 8
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK1-NEXT: call void @__omp_outlined__6(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]], i64 [[TMP4]]) #[[ATTR3]] // CHECK1-NEXT: call void @__omp_outlined__6(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]], i64 [[TMP4]]) #[[ATTR3]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__6 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__6
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[C:%.*]], i64 noundef [[F:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[C:%.*]], i64 noundef [[F:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 58 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK1-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP10]] to ptr // CHECK1-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP10]] to ptr
// CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8 // CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP17]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP17]], align 8
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK1-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP12]] to ptr // CHECK1-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP12]] to ptr
// CHECK1-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8 // CHECK1-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__7, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__7, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 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: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 123 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK1-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 // CHECK1-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 // CHECK1-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK1-NEXT: call void @__omp_outlined__8(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]]) #[[ATTR3]] // CHECK1-NEXT: call void @__omp_outlined__8(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]]) #[[ATTR3]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__8 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__8
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[C:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[C:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 93 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 8 // CHECK1-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 8
// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK1-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP20]] to ptr // CHECK1-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP20]] to ptr
// CHECK1-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8 // CHECK1-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP27]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP27]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP28:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[TMP28]], align 4 // CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[TMP28]], align 4
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP29]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__9, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP29]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__9, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 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: [[TMP30:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK1-NEXT: [[TMP30:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK1-NEXT: [[TMP31:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8 // CHECK1-NEXT: [[TMP31:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
// CHECK1-NEXT: [[ADD14:%.*]] = add nsw i64 [[TMP30]], [[TMP31]] // CHECK1-NEXT: [[ADD14:%.*]] = add nsw i64 [[TMP30]], [[TMP31]]
// CHECK1-NEXT: store i64 [[ADD14]], ptr [[DOTOMP_IV]], align 8 // CHECK1-NEXT: store i64 [[ADD14]], ptr [[DOTOMP_IV]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8 // CHECK1-NEXT: [[TMP32:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
// CHECK1-NEXT: [[TMP33:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8 // CHECK1-NEXT: [[TMP33:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
▲ Show 20 Lines • Show All 174 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[V_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[V_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK1-NEXT: store ptr [[V]], ptr [[V_ADDR]], align 8 // CHECK1-NEXT: store ptr [[V]], ptr [[V_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 // CHECK1-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[V_ADDR]], align 8 // CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[V_ADDR]], align 8
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK1-NEXT: call void @__omp_outlined__10(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]], ptr [[TMP5]]) #[[ATTR3]] // CHECK1-NEXT: call void @__omp_outlined__10(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]], ptr [[TMP5]]) #[[ATTR3]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__10 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__10
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[A:%.*]], ptr noundef [[V:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[A:%.*]], ptr noundef [[V:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 82 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP19]] to ptr // CHECK1-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP19]] to ptr
// CHECK1-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8 // CHECK1-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP27]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP27]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
// CHECK1-NEXT: store ptr [[TMP20]], ptr [[TMP28]], align 8 // CHECK1-NEXT: store ptr [[TMP20]], ptr [[TMP28]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4 // CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP30]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__11, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP30]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__11, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5)
// 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: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP31]], [[TMP32]] // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
// CHECK1-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 136 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[L_CASTED:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[L_CASTED:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK2-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK2-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK2-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK2-NEXT: store i64 [[L]], ptr [[L_ADDR]], align 8 // CHECK2-NEXT: store i64 [[L]], ptr [[L_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4:[0-9]+]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 // CHECK2-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8 // CHECK2-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[L_ADDR]], align 4 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[L_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP5]], ptr [[L_CASTED]], align 4 // CHECK2-NEXT: store i32 [[TMP5]], ptr [[L_CASTED]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = load i64, ptr [[L_CASTED]], align 8 // CHECK2-NEXT: [[TMP6:%.*]] = load i64, ptr [[L_CASTED]], align 8
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK2-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]], i64 [[TMP6]]) #[[ATTR3:[0-9]+]] // CHECK2-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]], i64 [[TMP6]]) #[[ATTR3:[0-9]+]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[A:%.*]], i64 noundef [[L:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[A:%.*]], i64 noundef [[L:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
▲ Show 20 Lines • Show All 86 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: store ptr [[TMP27]], ptr [[TMP26]], align 8 // CHECK2-NEXT: store ptr [[TMP27]], ptr [[TMP26]], align 8
// CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3 // CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP28]], align 8 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP28]], align 8
// CHECK2-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4 // CHECK2-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
// CHECK2-NEXT: [[TMP30:%.*]] = inttoptr i64 [[TMP21]] to ptr // CHECK2-NEXT: [[TMP30:%.*]] = inttoptr i64 [[TMP21]] to ptr
// CHECK2-NEXT: store ptr [[TMP30]], ptr [[TMP29]], align 8 // CHECK2-NEXT: store ptr [[TMP30]], ptr [[TMP29]], align 8
// CHECK2-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4 // CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP32]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP32]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5)
// 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: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP33]], [[TMP34]] // CHECK2-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP33]], [[TMP34]]
// CHECK2-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4 // CHECK2-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 177 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK2-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8 // CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 8 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 // CHECK2-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8 // CHECK2-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK2-NEXT: call void @__omp_outlined__2(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]]) #[[ATTR3]] // CHECK2-NEXT: call void @__omp_outlined__2(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]]) #[[ATTR3]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__2 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__2
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 2 dereferenceable(2000) [[AA:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 2 dereferenceable(2000) [[AA:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
▲ Show 20 Lines • Show All 77 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 8 // CHECK2-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 8
// CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2 // CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK2-NEXT: [[TMP25:%.*]] = inttoptr i64 [[TMP19]] to ptr // CHECK2-NEXT: [[TMP25:%.*]] = inttoptr i64 [[TMP19]] to ptr
// CHECK2-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 8 // CHECK2-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 8
// CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3 // CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 8 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 8
// CHECK2-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4 // CHECK2-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__3, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__3, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 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: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP29]], [[TMP30]] // CHECK2-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP29]], [[TMP30]]
// CHECK2-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 // CHECK2-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 127 Lines • ▼ Show 20 Lines
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l39 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l39
// CHECK2-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR0]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8 // CHECK2-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK2-NEXT: call void @__omp_outlined__4(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]]) #[[ATTR3]] // CHECK2-NEXT: call void @__omp_outlined__4(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]]) #[[ATTR3]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__4 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__4
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
▲ Show 20 Lines • Show All 46 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK2-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP8]] to ptr // CHECK2-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP8]] to ptr
// CHECK2-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8 // CHECK2-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8
// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 // CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK2-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP10]] to ptr // CHECK2-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP10]] to ptr
// CHECK2-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8 // CHECK2-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8
// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2 // CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 8 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 8
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__5, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__5, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3)
// 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: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]] // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK2-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 // CHECK2-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 100 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[F_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[F_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[F_CASTED:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[F_CASTED:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 // CHECK2-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8
// CHECK2-NEXT: store i64 [[F]], ptr [[F_ADDR]], align 8 // CHECK2-NEXT: store i64 [[F]], ptr [[F_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[F_ADDR]], align 4 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[F_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP3]], ptr [[F_CASTED]], align 4 // CHECK2-NEXT: store i32 [[TMP3]], ptr [[F_CASTED]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i64, ptr [[F_CASTED]], align 8 // CHECK2-NEXT: [[TMP4:%.*]] = load i64, ptr [[F_CASTED]], align 8
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK2-NEXT: call void @__omp_outlined__6(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]], i64 [[TMP4]]) #[[ATTR3]] // CHECK2-NEXT: call void @__omp_outlined__6(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]], i64 [[TMP4]]) #[[ATTR3]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__6 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__6
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[C:%.*]], i64 noundef [[F:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[C:%.*]], i64 noundef [[F:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
▲ Show 20 Lines • Show All 58 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 // CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK2-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP10]] to ptr // CHECK2-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP10]] to ptr
// CHECK2-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8 // CHECK2-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8
// CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2 // CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP17]], align 8 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP17]], align 8
// CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3 // CHECK2-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK2-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP12]] to ptr // CHECK2-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP12]] to ptr
// CHECK2-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8 // CHECK2-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__7, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__7, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 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: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]] // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
// CHECK2-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 // CHECK2-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 123 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK2-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK2-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 // CHECK2-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 // CHECK2-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8 // CHECK2-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK2-NEXT: call void @__omp_outlined__8(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]]) #[[ATTR3]] // CHECK2-NEXT: call void @__omp_outlined__8(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]]) #[[ATTR3]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__8 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__8
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[C:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[C:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
▲ Show 20 Lines • Show All 92 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8 // CHECK2-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8
// CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2 // CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK2-NEXT: [[TMP28:%.*]] = inttoptr i64 [[TMP22]] to ptr // CHECK2-NEXT: [[TMP28:%.*]] = inttoptr i64 [[TMP22]] to ptr
// CHECK2-NEXT: store ptr [[TMP28]], ptr [[TMP27]], align 8 // CHECK2-NEXT: store ptr [[TMP28]], ptr [[TMP27]], align 8
// CHECK2-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3 // CHECK2-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP29]], align 8 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP29]], align 8
// CHECK2-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4 // CHECK2-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP31]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__9, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP31]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__9, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 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: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP32]], [[TMP33]] // CHECK2-NEXT: [[ADD12:%.*]] = add nsw i32 [[TMP32]], [[TMP33]]
// CHECK2-NEXT: store i32 [[ADD12]], ptr [[DOTOMP_IV]], align 4 // CHECK2-NEXT: store i32 [[ADD12]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 170 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[V_ADDR:%.*]] = alloca ptr, align 8 // CHECK2-NEXT: [[V_ADDR:%.*]] = alloca ptr, align 8
// CHECK2-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK2-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK2-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK2-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK2-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK2-NEXT: store ptr [[V]], ptr [[V_ADDR]], align 8 // CHECK2-NEXT: store ptr [[V]], ptr [[V_ADDR]], align 8
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 // CHECK2-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8 // CHECK2-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[V_ADDR]], align 8 // CHECK2-NEXT: [[TMP5:%.*]] = load ptr, ptr [[V_ADDR]], align 8
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK2-NEXT: call void @__omp_outlined__10(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]], ptr [[TMP5]]) #[[ATTR3]] // CHECK2-NEXT: call void @__omp_outlined__10(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]], ptr [[TMP5]]) #[[ATTR3]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__10 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__10
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[A:%.*]], ptr noundef [[V:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[A:%.*]], ptr noundef [[V:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
▲ Show 20 Lines • Show All 82 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP19]] to ptr // CHECK2-NEXT: [[TMP26:%.*]] = inttoptr i64 [[TMP19]] to ptr
// CHECK2-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8 // CHECK2-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 8
// CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3 // CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP27]], align 8 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP27]], align 8
// CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4 // CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
// CHECK2-NEXT: store ptr [[TMP20]], ptr [[TMP28]], align 8 // CHECK2-NEXT: store ptr [[TMP20]], ptr [[TMP28]], align 8
// CHECK2-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK2-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK2-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4 // CHECK2-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP30]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__11, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP30]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__11, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5)
// 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: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP31]], [[TMP32]] // CHECK2-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
// CHECK2-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 // CHECK2-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 136 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[L_CASTED:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[L_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 // CHECK3-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
// CHECK3-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 // CHECK3-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK3-NEXT: store i32 [[L]], ptr [[L_ADDR]], align 4 // CHECK3-NEXT: store i32 [[L]], ptr [[L_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK3-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry: // CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4:[0-9]+]]) // CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 // CHECK3-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[N_CASTED]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[N_CASTED]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[L_ADDR]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[L_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], ptr [[L_CASTED]], align 4 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[L_CASTED]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[L_CASTED]], align 4 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[L_CASTED]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK3-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK3-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP4]], ptr [[TMP0]], i32 [[TMP6]]) #[[ATTR3:[0-9]+]] // CHECK3-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP4]], ptr [[TMP0]], i32 [[TMP6]]) #[[ATTR3:[0-9]+]]
// CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// CHECK3: worker.exit: // CHECK3: worker.exit:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[A:%.*]], i32 noundef [[L:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[A:%.*]], i32 noundef [[L:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
▲ Show 20 Lines • Show All 84 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 4 // CHECK3-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 4
// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 4
// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 4
// CHECK3-NEXT: [[TMP28:%.*]] = inttoptr i32 [[TMP19]] to ptr // CHECK3-NEXT: [[TMP28:%.*]] = inttoptr i32 [[TMP19]] to ptr
// CHECK3-NEXT: store ptr [[TMP28]], ptr [[TMP27]], align 4 // CHECK3-NEXT: store ptr [[TMP28]], ptr [[TMP27]], align 4
// CHECK3-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK3-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4 // CHECK3-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4
// CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP30]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 5) // CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP30]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 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: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK3-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP31]], [[TMP32]] // CHECK3-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
// CHECK3-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK3-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK3-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 172 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 // CHECK3-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
// CHECK3-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4 // CHECK3-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK3-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry: // CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 // CHECK3-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[N_CASTED]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[N_CASTED]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK3-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK3-NEXT: call void @__omp_outlined__2(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP4]], ptr [[TMP0]]) #[[ATTR3]] // CHECK3-NEXT: call void @__omp_outlined__2(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP4]], ptr [[TMP0]]) #[[ATTR3]]
// CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// CHECK3: worker.exit: // CHECK3: worker.exit:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__2 // CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__2
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[N:%.*]], ptr noundef nonnull align 2 dereferenceable(2000) [[AA:%.*]]) #[[ATTR1]] { // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[N:%.*]], ptr noundef nonnull align 2 dereferenceable(2000) [[AA:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
▲ Show 20 Lines • Show All 75 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: store ptr [[TMP21]], ptr [[TMP20]], align 4 // CHECK3-NEXT: store ptr [[TMP21]], ptr [[TMP20]], align 4
// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP23:%.*]] = inttoptr i32 [[TMP17]] to ptr // CHECK3-NEXT: [[TMP23:%.*]] = inttoptr i32 [[TMP17]] to ptr
// CHECK3-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 4 // CHECK3-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 4
// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP24]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP24]], align 4
// CHECK3-NEXT: [[TMP25:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK3-NEXT: [[TMP25:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4 // CHECK3-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
// CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP26]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__3, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4) // CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP26]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__3, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4)
// 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: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK3-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP27]], [[TMP28]] // CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP27]], [[TMP28]]
// CHECK3-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK3-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK3-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 123 Lines • ▼ Show 20 Lines
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l39 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l39
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR0]] { // CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR0]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 4 // CHECK3-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK3-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry: // CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK3-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK3-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK3-NEXT: call void @__omp_outlined__4(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]]) #[[ATTR3]] // CHECK3-NEXT: call void @__omp_outlined__4(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]]) #[[ATTR3]]
// CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// CHECK3: worker.exit: // CHECK3: worker.exit:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__4 // CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__4
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] { // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP10:%.*]] = inttoptr i32 [[TMP7]] to ptr // CHECK3-NEXT: [[TMP10:%.*]] = inttoptr i32 [[TMP7]] to ptr
// CHECK3-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 4 // CHECK3-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK3-NEXT: [[TMP12:%.*]] = inttoptr i32 [[TMP8]] to ptr // CHECK3-NEXT: [[TMP12:%.*]] = inttoptr i32 [[TMP8]] to ptr
// CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4 // CHECK3-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP13]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP13]], align 4
// CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__5, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3) // CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__5, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3)
// 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: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 96 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[F_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[F_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[F_CASTED:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[F_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 4 // CHECK3-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 4
// CHECK3-NEXT: store i32 [[F]], ptr [[F_ADDR]], align 4 // CHECK3-NEXT: store i32 [[F]], ptr [[F_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK3-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry: // CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[F_ADDR]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[F_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP3]], ptr [[F_CASTED]], align 4 // CHECK3-NEXT: store i32 [[TMP3]], ptr [[F_CASTED]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[F_CASTED]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[F_CASTED]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK3-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK3-NEXT: call void @__omp_outlined__6(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]], i32 [[TMP4]]) #[[ATTR3]] // CHECK3-NEXT: call void @__omp_outlined__6(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]], i32 [[TMP4]]) #[[ATTR3]]
// CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// CHECK3: worker.exit: // CHECK3: worker.exit:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__6 // CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__6
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[C:%.*]], i32 noundef [[F:%.*]]) #[[ATTR1]] { // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[C:%.*]], i32 noundef [[F:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
▲ Show 20 Lines • Show All 56 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK3-NEXT: [[TMP14:%.*]] = inttoptr i32 [[TMP8]] to ptr // CHECK3-NEXT: [[TMP14:%.*]] = inttoptr i32 [[TMP8]] to ptr
// CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 4 // CHECK3-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
// CHECK3-NEXT: [[TMP17:%.*]] = inttoptr i32 [[TMP10]] to ptr // CHECK3-NEXT: [[TMP17:%.*]] = inttoptr i32 [[TMP10]] to ptr
// CHECK3-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 4 // CHECK3-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 4
// CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__7, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4) // CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__7, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4)
// 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: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK3-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 118 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 // CHECK3-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
// CHECK3-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 4 // CHECK3-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK3-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry: // CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 // CHECK3-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[N_CASTED]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[N_CASTED]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK3-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK3-NEXT: call void @__omp_outlined__8(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP4]], ptr [[TMP0]]) #[[ATTR3]] // CHECK3-NEXT: call void @__omp_outlined__8(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP4]], ptr [[TMP0]]) #[[ATTR3]]
// CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// CHECK3: worker.exit: // CHECK3: worker.exit:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__8 // CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__8
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[C:%.*]]) #[[ATTR1]] { // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[C:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
▲ Show 20 Lines • Show All 95 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 4 // CHECK3-NEXT: store ptr [[TMP26]], ptr [[TMP25]], align 4
// CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
// CHECK3-NEXT: [[TMP28:%.*]] = inttoptr i32 [[TMP22]] to ptr // CHECK3-NEXT: [[TMP28:%.*]] = inttoptr i32 [[TMP22]] to ptr
// CHECK3-NEXT: store ptr [[TMP28]], ptr [[TMP27]], align 4 // CHECK3-NEXT: store ptr [[TMP28]], ptr [[TMP27]], align 4
// CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP29]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP29]], align 4
// CHECK3-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK3-NEXT: [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4 // CHECK3-NEXT: [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
// CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP31]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__9, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4) // CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP31]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__9, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4)
// 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: [[TMP32:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK3-NEXT: [[TMP32:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK3-NEXT: [[TMP33:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8 // CHECK3-NEXT: [[TMP33:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
// CHECK3-NEXT: [[ADD14:%.*]] = add nsw i64 [[TMP32]], [[TMP33]] // CHECK3-NEXT: [[ADD14:%.*]] = add nsw i64 [[TMP32]], [[TMP33]]
// CHECK3-NEXT: store i64 [[ADD14]], ptr [[DOTOMP_IV]], align 8 // CHECK3-NEXT: store i64 [[ADD14]], ptr [[DOTOMP_IV]], align 8
// CHECK3-NEXT: [[TMP34:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8 // CHECK3-NEXT: [[TMP34:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
// CHECK3-NEXT: [[TMP35:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8 // CHECK3-NEXT: [[TMP35:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
▲ Show 20 Lines • Show All 175 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[V_ADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[V_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 // CHECK3-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
// CHECK3-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 // CHECK3-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK3-NEXT: store ptr [[V]], ptr [[V_ADDR]], align 4 // CHECK3-NEXT: store ptr [[V]], ptr [[V_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK3-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry: // CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 // CHECK3-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[N_CASTED]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[N_CASTED]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[V_ADDR]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[V_ADDR]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK3-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK3-NEXT: call void @__omp_outlined__10(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP4]], ptr [[TMP0]], ptr [[TMP5]]) #[[ATTR3]] // CHECK3-NEXT: call void @__omp_outlined__10(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP4]], ptr [[TMP0]], ptr [[TMP5]]) #[[ATTR3]]
// CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// CHECK3: worker.exit: // CHECK3: worker.exit:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__10 // CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__10
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[A:%.*]], ptr noundef [[V:%.*]]) #[[ATTR1]] { // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[A:%.*]], ptr noundef [[V:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
▲ Show 20 Lines • Show All 80 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[TMP24:%.*]] = inttoptr i32 [[TMP17]] to ptr // CHECK3-NEXT: [[TMP24:%.*]] = inttoptr i32 [[TMP17]] to ptr
// CHECK3-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 4 // CHECK3-NEXT: store ptr [[TMP24]], ptr [[TMP23]], align 4
// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP25]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[TMP25]], align 4
// CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 4
// CHECK3-NEXT: store ptr [[TMP18]], ptr [[TMP26]], align 4 // CHECK3-NEXT: store ptr [[TMP18]], ptr [[TMP26]], align 4
// CHECK3-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK3-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4 // CHECK3-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
// CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__11, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 5) // CHECK3-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__11, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 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: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK3-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP29]], [[TMP30]] // CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP29]], [[TMP30]]
// CHECK3-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK3-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK3-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 124 Lines • Show Last 20 Lines

clang/test/OpenMP/nvptx_target_teams_distribute_parallel_for_generic_mode_codegen.cpp

Show All 37 Lines
// CHECK1-NEXT: [[ARGC_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[ARGC_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i64 [[ARGC]], ptr [[ARGC_ADDR]], align 8 // CHECK1-NEXT: store i64 [[ARGC]], ptr [[ARGC_ADDR]], align 8
// CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8 // CHECK1-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4:[0-9]+]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP3]], ptr [[ARGC_CASTED]], align 4 // CHECK1-NEXT: store i32 [[TMP3]], ptr [[ARGC_CASTED]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[ARGC_CASTED]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[ARGC_CASTED]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK1-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]], i64 [[TMP6]]) #[[ATTR3:[0-9]+]] // CHECK1-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]], i64 [[TMP6]]) #[[ATTR3:[0-9]+]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 86 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store ptr [[TMP27]], ptr [[TMP26]], align 8 // CHECK1-NEXT: store ptr [[TMP27]], ptr [[TMP26]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP28]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP28]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
// CHECK1-NEXT: [[TMP30:%.*]] = inttoptr i64 [[TMP21]] to ptr // CHECK1-NEXT: [[TMP30:%.*]] = inttoptr i64 [[TMP21]] to ptr
// CHECK1-NEXT: store ptr [[TMP30]], ptr [[TMP29]], align 8 // CHECK1-NEXT: store ptr [[TMP30]], ptr [[TMP29]], align 8
// CHECK1-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4 // CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP32]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5) // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP32]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5)
// 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: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP33]], [[TMP34]] // CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP33]], [[TMP34]]
// CHECK1-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK1-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 164 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[ARGC_CASTED:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[ARGC_CASTED:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK2-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK2-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 // CHECK2-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK2-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK2-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4:[0-9]+]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP3]], ptr [[ARGC_CASTED]], align 4 // CHECK2-NEXT: store i32 [[TMP3]], ptr [[ARGC_CASTED]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARGC_CASTED]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[ARGC_CASTED]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP5]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK2-NEXT: store i32 [[TMP5]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK2-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP4]], ptr [[TMP0]], i32 [[TMP6]]) #[[ATTR3:[0-9]+]] // CHECK2-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP4]], ptr [[TMP0]], i32 [[TMP6]]) #[[ATTR3:[0-9]+]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[ARGC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
▲ Show 20 Lines • Show All 84 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 4 // CHECK2-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 4
// CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3 // CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 4 // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 4
// CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 4 // CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 4
// CHECK2-NEXT: [[TMP28:%.*]] = inttoptr i32 [[TMP19]] to ptr // CHECK2-NEXT: [[TMP28:%.*]] = inttoptr i32 [[TMP19]] to ptr
// CHECK2-NEXT: store ptr [[TMP28]], ptr [[TMP27]], align 4 // CHECK2-NEXT: store ptr [[TMP28]], ptr [[TMP27]], align 4
// CHECK2-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK2-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK2-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4 // CHECK2-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP30]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 5) // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP30]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 5)
// 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: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 // CHECK2-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
// CHECK2-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP31]], [[TMP32]] // CHECK2-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
// CHECK2-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4 // CHECK2-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
// CHECK2-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK2-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 // CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4
▲ Show 20 Lines • Show All 153 Lines • Show Last 20 Lines

clang/test/OpenMP/nvptx_target_teams_distribute_parallel_for_simd_codegen.cpp

Show First 20 Lines • Show All 67 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[L_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[L_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK1-NEXT: store i64 [[L]], ptr [[L_ADDR]], align 8 // CHECK1-NEXT: store i64 [[L]], ptr [[L_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4:[0-9]+]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 // CHECK1-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[L_ADDR]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[L_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], ptr [[L_CASTED]], align 4 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[L_CASTED]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[L_CASTED]], align 8 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[L_CASTED]], align 8
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK1-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]], i64 [[TMP6]]) #[[ATTR3:[0-9]+]] // CHECK1-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]], i64 [[TMP6]]) #[[ATTR3:[0-9]+]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[A:%.*]], i64 noundef [[L:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[A:%.*]], i64 noundef [[L:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 86 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store ptr [[TMP27]], ptr [[TMP26]], align 8, !llvm.access.group [[ACC_GRP12]] // CHECK1-NEXT: store ptr [[TMP27]], ptr [[TMP26]], align 8, !llvm.access.group [[ACC_GRP12]]
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP28]], align 8, !llvm.access.group [[ACC_GRP12]] // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP28]], align 8, !llvm.access.group [[ACC_GRP12]]
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 4
// CHECK1-NEXT: [[TMP30:%.*]] = inttoptr i64 [[TMP21]] to ptr // CHECK1-NEXT: [[TMP30:%.*]] = inttoptr i64 [[TMP21]] to ptr
// CHECK1-NEXT: store ptr [[TMP30]], ptr [[TMP29]], align 8, !llvm.access.group [[ACC_GRP12]] // CHECK1-NEXT: store ptr [[TMP30]], ptr [[TMP29]], align 8, !llvm.access.group [[ACC_GRP12]]
// CHECK1-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8, !llvm.access.group [[ACC_GRP12]] // CHECK1-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8, !llvm.access.group [[ACC_GRP12]]
// CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4, !llvm.access.group [[ACC_GRP12]] // CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP32]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5), !llvm.access.group [[ACC_GRP12]] // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP32]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 5), !llvm.access.group [[ACC_GRP12]]
// 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: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]] // CHECK1-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK1-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP12]] // CHECK1-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP33]], [[TMP34]] // CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP33]], [[TMP34]]
// CHECK1-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]] // CHECK1-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP12]] // CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK1-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP12]] // CHECK1-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP12]]
▲ Show 20 Lines • Show All 201 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK1-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 // CHECK1-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK1-NEXT: call void @__omp_outlined__2(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]]) #[[ATTR3]] // CHECK1-NEXT: call void @__omp_outlined__2(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i64 [[TMP4]], ptr [[TMP0]]) #[[ATTR3]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__2 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__2
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 2 dereferenceable(2000) [[AA:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 2 dereferenceable(2000) [[AA:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 77 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 8, !llvm.access.group [[ACC_GRP19]] // CHECK1-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 8, !llvm.access.group [[ACC_GRP19]]
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK1-NEXT: [[TMP25:%.*]] = inttoptr i64 [[TMP19]] to ptr // CHECK1-NEXT: [[TMP25:%.*]] = inttoptr i64 [[TMP19]] to ptr
// CHECK1-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 8, !llvm.access.group [[ACC_GRP19]] // CHECK1-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 8, !llvm.access.group [[ACC_GRP19]]
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 8, !llvm.access.group [[ACC_GRP19]] // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 8, !llvm.access.group [[ACC_GRP19]]
// CHECK1-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8, !llvm.access.group [[ACC_GRP19]] // CHECK1-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8, !llvm.access.group [[ACC_GRP19]]
// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4, !llvm.access.group [[ACC_GRP19]] // CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4, !llvm.access.group [[ACC_GRP19]]
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__3, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4), !llvm.access.group [[ACC_GRP19]] // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP28]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__3, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4), !llvm.access.group [[ACC_GRP19]]
// 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: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP19]] // CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP19]]
// CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP19]] // CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP19]]
// CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP29]], [[TMP30]] // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP29]], [[TMP30]]
// CHECK1-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP19]] // CHECK1-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP19]]
// CHECK1-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP19]] // CHECK1-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP19]]
// CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP19]] // CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP19]]
▲ Show 20 Lines • Show All 151 Lines • ▼ Show 20 Lines
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l37 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l37
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8 // CHECK1-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK1-NEXT: call void @__omp_outlined__4(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]]) #[[ATTR3]] // CHECK1-NEXT: call void @__omp_outlined__4(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]]) #[[ATTR3]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__4 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__4
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 46 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK1-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP8]] to ptr // CHECK1-NEXT: [[TMP12:%.*]] = inttoptr i64 [[TMP8]] to ptr
// CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8, !llvm.access.group [[ACC_GRP25]] // CHECK1-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 8, !llvm.access.group [[ACC_GRP25]]
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK1-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP10]] to ptr // CHECK1-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP10]] to ptr
// CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8, !llvm.access.group [[ACC_GRP25]] // CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8, !llvm.access.group [[ACC_GRP25]]
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 8, !llvm.access.group [[ACC_GRP25]] // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 8, !llvm.access.group [[ACC_GRP25]]
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__5, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3), !llvm.access.group [[ACC_GRP25]] // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__5, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 3), !llvm.access.group [[ACC_GRP25]]
// 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: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP25]] // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP25]]
// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP25]] // CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP25]]
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP25]] // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP25]]
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP25]] // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP25]]
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP25]] // CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP25]]
▲ Show 20 Lines • Show All 114 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[F_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[F_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[F_CASTED:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[F_CASTED:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8 // CHECK1-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 8
// CHECK1-NEXT: store i64 [[F]], ptr [[F_ADDR]], align 8 // CHECK1-NEXT: store i64 [[F]], ptr [[F_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK1-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[F_ADDR]], align 4 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[F_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP3]], ptr [[F_CASTED]], align 4 // CHECK1-NEXT: store i32 [[TMP3]], ptr [[F_CASTED]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[F_CASTED]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[F_CASTED]], align 8
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK1-NEXT: call void @__omp_outlined__6(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]], i64 [[TMP4]]) #[[ATTR3]] // CHECK1-NEXT: call void @__omp_outlined__6(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]], i64 [[TMP4]]) #[[ATTR3]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__6 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__6
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[C:%.*]], i64 noundef [[F:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[C:%.*]], i64 noundef [[F:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 58 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1
// CHECK1-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP10]] to ptr // CHECK1-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP10]] to ptr
// CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8, !llvm.access.group [[ACC_GRP31]] // CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8, !llvm.access.group [[ACC_GRP31]]
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP17]], align 8, !llvm.access.group [[ACC_GRP31]] // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP17]], align 8, !llvm.access.group [[ACC_GRP31]]
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3
// CHECK1-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP12]] to ptr // CHECK1-NEXT: [[TMP19:%.*]] = inttoptr i64 [[TMP12]] to ptr
// CHECK1-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8, !llvm.access.group [[ACC_GRP31]] // CHECK1-NEXT: store ptr [[TMP19]], ptr [[TMP18]], align 8, !llvm.access.group [[ACC_GRP31]]
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__7, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4), !llvm.access.group [[ACC_GRP31]] // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__7, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4), !llvm.access.group [[ACC_GRP31]]
// 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: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP31]] // CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP31]]
// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP31]] // CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP31]]
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP31]] // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP31]]
// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP31]] // CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP31]]
// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP31]] // CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP31]]
▲ Show 20 Lines • Show All 142 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[L_CASTED:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[L_CASTED:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 // CHECK2-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
// CHECK2-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 // CHECK2-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4
// CHECK2-NEXT: store i32 [[L]], ptr [[L_ADDR]], align 4 // CHECK2-NEXT: store i32 [[L]], ptr [[L_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4:[0-9]+]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 // CHECK2-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[N_CASTED]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[N_CASTED]], align 4
// CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[L_ADDR]], align 4 // CHECK2-NEXT: [[TMP5:%.*]] = load i32, ptr [[L_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP5]], ptr [[L_CASTED]], align 4 // CHECK2-NEXT: store i32 [[TMP5]], ptr [[L_CASTED]], align 4
// CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[L_CASTED]], align 4 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, ptr [[L_CASTED]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK2-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP4]], ptr [[TMP0]], i32 [[TMP6]]) #[[ATTR3:[0-9]+]] // CHECK2-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP4]], ptr [[TMP0]], i32 [[TMP6]]) #[[ATTR3:[0-9]+]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[A:%.*]], i32 noundef [[L:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[A:%.*]], i32 noundef [[L:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
▲ Show 20 Lines • Show All 84 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 4, !llvm.access.group [[ACC_GRP12]] // CHECK2-NEXT: store ptr [[TMP25]], ptr [[TMP24]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3 // CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 4, !llvm.access.group [[ACC_GRP12]] // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 4 // CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 4
// CHECK2-NEXT: [[TMP28:%.*]] = inttoptr i32 [[TMP19]] to ptr // CHECK2-NEXT: [[TMP28:%.*]] = inttoptr i32 [[TMP19]] to ptr
// CHECK2-NEXT: store ptr [[TMP28]], ptr [[TMP27]], align 4, !llvm.access.group [[ACC_GRP12]] // CHECK2-NEXT: store ptr [[TMP28]], ptr [[TMP27]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK2-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4, !llvm.access.group [[ACC_GRP12]] // CHECK2-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK2-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4, !llvm.access.group [[ACC_GRP12]] // CHECK2-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP30]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 5), !llvm.access.group [[ACC_GRP12]] // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP30]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__1, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 5), !llvm.access.group [[ACC_GRP12]]
// 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: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]] // CHECK2-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP12]] // CHECK2-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK2-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP31]], [[TMP32]] // CHECK2-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP31]], [[TMP32]]
// CHECK2-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]] // CHECK2-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK2-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP12]] // CHECK2-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP12]] // CHECK2-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP12]]
▲ Show 20 Lines • Show All 196 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[AA_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 // CHECK2-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4
// CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: store ptr [[AA]], ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[AA_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 // CHECK2-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[N_CASTED]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[N_CASTED]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK2-NEXT: call void @__omp_outlined__2(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP4]], ptr [[TMP0]]) #[[ATTR3]] // CHECK2-NEXT: call void @__omp_outlined__2(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], i32 [[TMP4]], ptr [[TMP0]]) #[[ATTR3]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__2 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__2
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[N:%.*]], ptr noundef nonnull align 2 dereferenceable(2000) [[AA:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[N:%.*]], ptr noundef nonnull align 2 dereferenceable(2000) [[AA:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
▲ Show 20 Lines • Show All 75 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: store ptr [[TMP21]], ptr [[TMP20]], align 4, !llvm.access.group [[ACC_GRP19]] // CHECK2-NEXT: store ptr [[TMP21]], ptr [[TMP20]], align 4, !llvm.access.group [[ACC_GRP19]]
// CHECK2-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2 // CHECK2-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
// CHECK2-NEXT: [[TMP23:%.*]] = inttoptr i32 [[TMP17]] to ptr // CHECK2-NEXT: [[TMP23:%.*]] = inttoptr i32 [[TMP17]] to ptr
// CHECK2-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 4, !llvm.access.group [[ACC_GRP19]] // CHECK2-NEXT: store ptr [[TMP23]], ptr [[TMP22]], align 4, !llvm.access.group [[ACC_GRP19]]
// CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3 // CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP24]], align 4, !llvm.access.group [[ACC_GRP19]] // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP24]], align 4, !llvm.access.group [[ACC_GRP19]]
// CHECK2-NEXT: [[TMP25:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4, !llvm.access.group [[ACC_GRP19]] // CHECK2-NEXT: [[TMP25:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4, !llvm.access.group [[ACC_GRP19]]
// CHECK2-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4, !llvm.access.group [[ACC_GRP19]] // CHECK2-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4, !llvm.access.group [[ACC_GRP19]]
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP26]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__3, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4), !llvm.access.group [[ACC_GRP19]] // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP26]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__3, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4), !llvm.access.group [[ACC_GRP19]]
// 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: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP19]] // CHECK2-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP19]]
// CHECK2-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP19]] // CHECK2-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP19]]
// CHECK2-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP27]], [[TMP28]] // CHECK2-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP27]], [[TMP28]]
// CHECK2-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP19]] // CHECK2-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP19]]
// CHECK2-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP19]] // CHECK2-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP19]]
// CHECK2-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP19]] // CHECK2-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP19]]
▲ Show 20 Lines • Show All 147 Lines • ▼ Show 20 Lines
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l37 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9ftemplateIiET_i_l37
// CHECK2-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR0]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 4 // CHECK2-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[B_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK2-NEXT: call void @__omp_outlined__4(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]]) #[[ATTR3]] // CHECK2-NEXT: call void @__omp_outlined__4(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]]) #[[ATTR3]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__4 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__4
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[B:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
▲ Show 20 Lines • Show All 44 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0 // CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 0
// CHECK2-NEXT: [[TMP10:%.*]] = inttoptr i32 [[TMP7]] to ptr // CHECK2-NEXT: [[TMP10:%.*]] = inttoptr i32 [[TMP7]] to ptr
// CHECK2-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 4, !llvm.access.group [[ACC_GRP25]] // CHECK2-NEXT: store ptr [[TMP10]], ptr [[TMP9]], align 4, !llvm.access.group [[ACC_GRP25]]
// CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1 // CHECK2-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK2-NEXT: [[TMP12:%.*]] = inttoptr i32 [[TMP8]] to ptr // CHECK2-NEXT: [[TMP12:%.*]] = inttoptr i32 [[TMP8]] to ptr
// CHECK2-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4, !llvm.access.group [[ACC_GRP25]] // CHECK2-NEXT: store ptr [[TMP12]], ptr [[TMP11]], align 4, !llvm.access.group [[ACC_GRP25]]
// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2 // CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP13]], align 4, !llvm.access.group [[ACC_GRP25]] // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP13]], align 4, !llvm.access.group [[ACC_GRP25]]
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__5, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3), !llvm.access.group [[ACC_GRP25]] // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__5, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 3), !llvm.access.group [[ACC_GRP25]]
// 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: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP25]] // CHECK2-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP25]]
// CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP25]] // CHECK2-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP25]]
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK2-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP25]] // CHECK2-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP25]]
// CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP25]] // CHECK2-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP25]]
// CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP25]] // CHECK2-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP25]]
▲ Show 20 Lines • Show All 110 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[C_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[F_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[F_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[F_CASTED:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[F_CASTED:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 4 // CHECK2-NEXT: store ptr [[C]], ptr [[C_ADDR]], align 4
// CHECK2-NEXT: store i32 [[F]], ptr [[F_ADDR]], align 4 // CHECK2-NEXT: store i32 [[F]], ptr [[F_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 4 // CHECK2-NEXT: [[TMP0:%.*]] = load ptr, ptr [[C_ADDR]], align 4
// CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false, i1 true) // CHECK2-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 2, i1 false)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP1]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB4]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[F_ADDR]], align 4 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, ptr [[F_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP3]], ptr [[F_CASTED]], align 4 // CHECK2-NEXT: store i32 [[TMP3]], ptr [[F_CASTED]], align 4
// CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[F_CASTED]], align 4 // CHECK2-NEXT: [[TMP4:%.*]] = load i32, ptr [[F_CASTED]], align 4
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK2-NEXT: call void @__omp_outlined__6(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]], i32 [[TMP4]]) #[[ATTR3]] // CHECK2-NEXT: call void @__omp_outlined__6(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[TMP0]], i32 [[TMP4]]) #[[ATTR3]]
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__6 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__6
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[C:%.*]], i32 noundef [[F:%.*]]) #[[ATTR1]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[C:%.*]], i32 noundef [[F:%.*]]) #[[ATTR1]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
▲ Show 20 Lines • Show All 56 Lines • ▼ Show 20 Lines
// CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1 // CHECK2-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 1
// CHECK2-NEXT: [[TMP14:%.*]] = inttoptr i32 [[TMP8]] to ptr // CHECK2-NEXT: [[TMP14:%.*]] = inttoptr i32 [[TMP8]] to ptr
// CHECK2-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 4, !llvm.access.group [[ACC_GRP31]] // CHECK2-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 4, !llvm.access.group [[ACC_GRP31]]
// CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2 // CHECK2-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 2
// CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 4, !llvm.access.group [[ACC_GRP31]] // CHECK2-NEXT: store ptr [[TMP0]], ptr [[TMP15]], align 4, !llvm.access.group [[ACC_GRP31]]
// CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3 // CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i32 0, i32 3
// CHECK2-NEXT: [[TMP17:%.*]] = inttoptr i32 [[TMP10]] to ptr // CHECK2-NEXT: [[TMP17:%.*]] = inttoptr i32 [[TMP10]] to ptr
// CHECK2-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 4, !llvm.access.group [[ACC_GRP31]] // CHECK2-NEXT: store ptr [[TMP17]], ptr [[TMP16]], align 4, !llvm.access.group [[ACC_GRP31]]
// CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB4]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__7, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4), !llvm.access.group [[ACC_GRP31]] // CHECK2-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB1]], i32 [[TMP2]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__7, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i32 4), !llvm.access.group [[ACC_GRP31]]
// 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: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP31]] // CHECK2-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP31]]
// CHECK2-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP31]] // CHECK2-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP31]]
// CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
// CHECK2-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP31]] // CHECK2-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP31]]
// CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP31]] // CHECK2-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP31]]
// CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP31]] // CHECK2-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP31]]
▲ Show 20 Lines • Show All 130 Lines • Show Last 20 Lines

clang/test/OpenMP/nvptx_target_teams_distribute_simd_codegen.cpp

Show First 20 Lines • Show All 64 Lines • ▼ Show 20 Linesint bar(int n){
int a = 0; int a = 0;
a += ftemplate<int>(n); a += ftemplate<int>(n);
return a; return a;
} }
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}_l37( // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}_l37(
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2, i1 true) // CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2)
// CHECK: call void @__kmpc_distribute_static_init_4({{.+}}, {{.+}}, {{.+}} 91, // CHECK: call void @__kmpc_distribute_static_init_4({{.+}}, {{.+}}, {{.+}} 91,
// CHECK: call void @__kmpc_distribute_static_fini( // CHECK: call void @__kmpc_distribute_static_fini(
// CHECK: ret void // CHECK: ret void
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}_l43( // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}_l43(
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2, i1 true) // CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2)
// CHECK: call void @__kmpc_distribute_static_init_4({{.+}}, {{.+}}, {{.+}} 91, // CHECK: call void @__kmpc_distribute_static_init_4({{.+}}, {{.+}}, {{.+}} 91,
// CHECK: call void @__kmpc_distribute_static_fini( // CHECK: call void @__kmpc_distribute_static_fini(
// CHECK: ret void // CHECK: ret void
// CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}_l48( // CHECK-LABEL: define {{.*}}void {{@__omp_offloading_.+}}_l48(
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2, i1 true) // CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2)
// CHECK: call void @__kmpc_distribute_static_init_4({{.+}}, {{.+}}, {{.+}} 91, // CHECK: call void @__kmpc_distribute_static_init_4({{.+}}, {{.+}}, {{.+}} 91,
// CHECK: call void @__kmpc_distribute_static_fini( // CHECK: call void @__kmpc_distribute_static_fini(
// CHECK: ret void // CHECK: ret void
// CHECK: define {{.*}}void {{@__omp_offloading_.+}}_l53({{.+}}, i{{32|64}} [[F_IN:%.+]]) // CHECK: define {{.*}}void {{@__omp_offloading_.+}}_l53({{.+}}, i{{32|64}} [[F_IN:%.+]])
// CHECK: store {{.+}} [[F_IN]], ptr {{.+}}, // CHECK: store {{.+}} [[F_IN]], ptr {{.+}},
// CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false, i1 true) // CHECK: call i32 @__kmpc_target_init({{.*}}, i8 2, i1 false)
// CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2, i1 true) // CHECK: call void @__kmpc_target_deinit({{.*}}, i8 2)
// CHECK: store {{.+}} 99, ptr [[COMB_UB:%.+]], align // CHECK: store {{.+}} 99, ptr [[COMB_UB:%.+]], align
// CHECK: call void @__kmpc_distribute_static_init_4({{.+}}, {{.+}}, {{.+}} 91, {{.+}}, {{.+}}, ptr [[COMB_UB]], // CHECK: call void @__kmpc_distribute_static_init_4({{.+}}, {{.+}}, {{.+}} 91, {{.+}}, {{.+}}, ptr [[COMB_UB]],
// CHECK: call void @__kmpc_distribute_static_fini( // CHECK: call void @__kmpc_distribute_static_fini(
// CHECK: ret void // CHECK: ret void
#endif #endif

clang/test/OpenMP/nvptx_teams_codegen.cpp

Show First 20 Lines • Show All 77 Lines • ▼ Show 20 Lines
#endif #endif
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l23 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l23
// CHECK1-SAME: (i64 noundef [[ARGC:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK1-SAME: (i64 noundef [[ARGC:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8 // CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i64 [[ARGC]], ptr [[ARGC_ADDR]], align 8 // CHECK1-NEXT: store i64 [[ARGC]], ptr [[ARGC_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK1-NEXT: [[ARGC1:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 4) // CHECK1-NEXT: [[ARGC1:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 4)
// CHECK1-NEXT: store i32 [[TMP1]], ptr [[ARGC1]], align 4 // CHECK1-NEXT: store i32 [[TMP1]], ptr [[ARGC1]], align 4
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK1-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[ARGC1]]) #[[ATTR3:[0-9]+]] // CHECK1-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[ARGC1]]) #[[ATTR3:[0-9]+]]
// CHECK1-NEXT: call void @__kmpc_free_shared(ptr [[ARGC1]], i64 4) // CHECK1-NEXT: call void @__kmpc_free_shared(ptr [[ARGC1]], i64 4)
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
Show All 10 Lines
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPPcEiT__l15 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPPcEiT__l15
// CHECK1-SAME: (ptr noundef [[ARGC:%.*]]) #[[ATTR0]] { // CHECK1-SAME: (ptr noundef [[ARGC:%.*]]) #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store ptr [[ARGC]], ptr [[ARGC_ADDR]], align 8 // CHECK1-NEXT: store ptr [[ARGC]], ptr [[ARGC_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[ARGC_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[ARGC_ADDR]], align 8
// CHECK1-NEXT: [[ARGC1:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 8) // CHECK1-NEXT: [[ARGC1:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 8)
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[ARGC1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[ARGC1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK1-NEXT: call void @__omp_outlined__1(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[ARGC1]]) #[[ATTR3]] // CHECK1-NEXT: call void @__omp_outlined__1(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[ARGC1]]) #[[ATTR3]]
// CHECK1-NEXT: call void @__kmpc_free_shared(ptr [[ARGC1]], i64 8) // CHECK1-NEXT: call void @__kmpc_free_shared(ptr [[ARGC1]], i64 8)
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[ARGC:%.*]]) #[[ATTR2]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[ARGC:%.*]]) #[[ATTR2]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
Show All 10 Lines
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l23 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l23
// CHECK2-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK2-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK2-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK2-NEXT: [[ARGC1:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i32 4) // CHECK2-NEXT: [[ARGC1:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i32 4)
// CHECK2-NEXT: store i32 [[TMP1]], ptr [[ARGC1]], align 4 // CHECK2-NEXT: store i32 [[TMP1]], ptr [[ARGC1]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK2-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[ARGC1]]) #[[ATTR3:[0-9]+]] // CHECK2-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[ARGC1]]) #[[ATTR3:[0-9]+]]
// CHECK2-NEXT: call void @__kmpc_free_shared(ptr [[ARGC1]], i32 4) // CHECK2-NEXT: call void @__kmpc_free_shared(ptr [[ARGC1]], i32 4)
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 10 Lines
// //
// CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPPcEiT__l15 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPPcEiT__l15
// CHECK2-SAME: (ptr noundef [[ARGC:%.*]]) #[[ATTR0]] { // CHECK2-SAME: (ptr noundef [[ARGC:%.*]]) #[[ATTR0]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
// CHECK2-NEXT: [[ARGC_ADDR:%.*]] = alloca ptr, align 4 // CHECK2-NEXT: [[ARGC_ADDR:%.*]] = alloca ptr, align 4
// CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK2-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK2-NEXT: store ptr [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK2-NEXT: store ptr [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK2-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK2-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK2-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK2: user_code.entry: // CHECK2: user_code.entry:
// CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[ARGC_ADDR]], align 4 // CHECK2-NEXT: [[TMP1:%.*]] = load ptr, ptr [[ARGC_ADDR]], align 4
// CHECK2-NEXT: [[ARGC1:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i32 4) // CHECK2-NEXT: [[ARGC1:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i32 4)
// CHECK2-NEXT: store ptr [[TMP1]], ptr [[ARGC1]], align 4 // CHECK2-NEXT: store ptr [[TMP1]], ptr [[ARGC1]], align 4
// CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK2-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK2-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK2-NEXT: call void @__omp_outlined__1(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[ARGC1]]) #[[ATTR3]] // CHECK2-NEXT: call void @__omp_outlined__1(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[ARGC1]]) #[[ATTR3]]
// CHECK2-NEXT: call void @__kmpc_free_shared(ptr [[ARGC1]], i32 4) // CHECK2-NEXT: call void @__kmpc_free_shared(ptr [[ARGC1]], i32 4)
// CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK2-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// CHECK2: worker.exit: // CHECK2: worker.exit:
// CHECK2-NEXT: ret void // CHECK2-NEXT: ret void
// //
// //
// CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK2-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]]) #[[ATTR2]] { // CHECK2-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]]) #[[ATTR2]] {
// CHECK2-NEXT: entry: // CHECK2-NEXT: entry:
Show All 14 Lines
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK3-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8 // CHECK3-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8
// CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8 // CHECK3-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8
// CHECK3-NEXT: store i64 [[B]], ptr [[B_ADDR]], align 8 // CHECK3-NEXT: store i64 [[B]], ptr [[B_ADDR]], align 8
// CHECK3-NEXT: store i64 [[ARGC]], ptr [[ARGC_ADDR]], align 8 // CHECK3-NEXT: store i64 [[ARGC]], ptr [[ARGC_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry: // CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK3-NEXT: [[ARGC3:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 4) // CHECK3-NEXT: [[ARGC1:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 4)
// CHECK3-NEXT: store i32 [[TMP1]], ptr [[ARGC3]], align 4 // CHECK3-NEXT: store i32 [[TMP1]], ptr [[ARGC1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK3-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK3-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK3-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[ARGC3]]) #[[ATTR3:[0-9]+]] // CHECK3-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[ARGC1]]) #[[ATTR3:[0-9]+]]
// CHECK3-NEXT: call void @__kmpc_free_shared(ptr [[ARGC3]], i64 4) // CHECK3-NEXT: call void @__kmpc_free_shared(ptr [[ARGC1]], i64 4)
// CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// CHECK3: worker.exit: // CHECK3: worker.exit:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
Show All 14 Lines
// CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca i64, align 8
// CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8 // CHECK3-NEXT: [[B_ADDR:%.*]] = alloca i64, align 8
// CHECK3-NEXT: [[ARGC_ADDR:%.*]] = alloca ptr, align 8 // CHECK3-NEXT: [[ARGC_ADDR:%.*]] = alloca ptr, align 8
// CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8 // CHECK3-NEXT: store i64 [[A]], ptr [[A_ADDR]], align 8
// CHECK3-NEXT: store i64 [[B]], ptr [[B_ADDR]], align 8 // CHECK3-NEXT: store i64 [[B]], ptr [[B_ADDR]], align 8
// CHECK3-NEXT: store ptr [[ARGC]], ptr [[ARGC_ADDR]], align 8 // CHECK3-NEXT: store ptr [[ARGC]], ptr [[ARGC_ADDR]], align 8
// CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK3-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK3-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK3: user_code.entry: // CHECK3: user_code.entry:
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[ARGC_ADDR]], align 8 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[ARGC_ADDR]], align 8
// CHECK3-NEXT: [[ARGC2:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 8) // CHECK3-NEXT: [[ARGC1:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i64 8)
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[ARGC2]], align 8 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[ARGC1]], align 8
// CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK3-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK3-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK3-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK3-NEXT: call void @__omp_outlined__1(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[ARGC2]]) #[[ATTR3]] // CHECK3-NEXT: call void @__omp_outlined__1(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[ARGC1]]) #[[ATTR3]]
// CHECK3-NEXT: call void @__kmpc_free_shared(ptr [[ARGC2]], i64 8) // CHECK3-NEXT: call void @__kmpc_free_shared(ptr [[ARGC1]], i64 8)
// CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK3-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// CHECK3: worker.exit: // CHECK3: worker.exit:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK3-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[ARGC:%.*]]) #[[ATTR2]] { // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[ARGC:%.*]]) #[[ATTR2]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
Show All 14 Lines
// CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK4-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 // CHECK4-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK4-NEXT: store i32 [[B]], ptr [[B_ADDR]], align 4 // CHECK4-NEXT: store i32 [[B]], ptr [[B_ADDR]], align 4
// CHECK4-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK4-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK4-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) // CHECK4-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 1, i1 true)
// CHECK4-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK4-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK4-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK4-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK4: user_code.entry: // CHECK4: user_code.entry:
// CHECK4-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK4-NEXT: [[ARGC1:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i32 4) // CHECK4-NEXT: [[ARGC1:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i32 4)
// CHECK4-NEXT: store i32 [[TMP1]], ptr [[ARGC1]], align 4 // CHECK4-NEXT: store i32 [[TMP1]], ptr [[ARGC1]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK4-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK4-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK4-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK4-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK4-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK4-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[ARGC1]]) #[[ATTR3:[0-9]+]] // CHECK4-NEXT: call void @__omp_outlined__(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[ARGC1]]) #[[ATTR3:[0-9]+]]
// CHECK4-NEXT: call void @__kmpc_free_shared(ptr [[ARGC1]], i32 4) // CHECK4-NEXT: call void @__kmpc_free_shared(ptr [[ARGC1]], i32 4)
// CHECK4-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK4-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// CHECK4: worker.exit: // CHECK4: worker.exit:
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK4-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK4-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK4-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
Show All 14 Lines
// CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[B_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[ARGC_ADDR:%.*]] = alloca ptr, align 4 // CHECK4-NEXT: [[ARGC_ADDR:%.*]] = alloca ptr, align 4
// CHECK4-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
// CHECK4-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 // CHECK4-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
// CHECK4-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4 // CHECK4-NEXT: store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK4-NEXT: store i32 [[B]], ptr [[B_ADDR]], align 4 // CHECK4-NEXT: store i32 [[B]], ptr [[B_ADDR]], align 4
// CHECK4-NEXT: store ptr [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK4-NEXT: store ptr [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK4-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true, i1 true) // CHECK4-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1]], i8 1, i1 true)
// CHECK4-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // CHECK4-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
// CHECK4-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] // CHECK4-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK4: user_code.entry: // CHECK4: user_code.entry:
// CHECK4-NEXT: [[TMP1:%.*]] = load ptr, ptr [[ARGC_ADDR]], align 4 // CHECK4-NEXT: [[TMP1:%.*]] = load ptr, ptr [[ARGC_ADDR]], align 4
// CHECK4-NEXT: [[ARGC1:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i32 4) // CHECK4-NEXT: [[ARGC1:%.*]] = call align 8 ptr @__kmpc_alloc_shared(i32 4)
// CHECK4-NEXT: store ptr [[TMP1]], ptr [[ARGC1]], align 4 // CHECK4-NEXT: store ptr [[TMP1]], ptr [[ARGC1]], align 4
// CHECK4-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]]) // CHECK4-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB1]])
// CHECK4-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4 // CHECK4-NEXT: store i32 0, ptr [[DOTZERO_ADDR]], align 4
// CHECK4-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4 // CHECK4-NEXT: store i32 [[TMP2]], ptr [[DOTTHREADID_TEMP_]], align 4
// CHECK4-NEXT: call void @__omp_outlined__1(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[ARGC1]]) #[[ATTR3]] // CHECK4-NEXT: call void @__omp_outlined__1(ptr [[DOTTHREADID_TEMP_]], ptr [[DOTZERO_ADDR]], ptr [[ARGC1]]) #[[ATTR3]]
// CHECK4-NEXT: call void @__kmpc_free_shared(ptr [[ARGC1]], i32 4) // CHECK4-NEXT: call void @__kmpc_free_shared(ptr [[ARGC1]], i32 4)
// CHECK4-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1, i1 true) // CHECK4-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 1)
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// CHECK4: worker.exit: // CHECK4: worker.exit:
// CHECK4-NEXT: ret void // CHECK4-NEXT: ret void
// //
// //
// CHECK4-LABEL: define {{[^@]+}}@__omp_outlined__1 // CHECK4-LABEL: define {{[^@]+}}@__omp_outlined__1
// CHECK4-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]]) #[[ATTR2]] { // CHECK4-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]]) #[[ATTR2]] {
// CHECK4-NEXT: entry: // CHECK4-NEXT: entry:
Show All 10 Lines

clang/test/OpenMP/nvptx_teams_reduction_codegen.cpp

This file has a very large number of changes (3,012 lines). Show File Contents

clang/test/OpenMP/reduction_implicit_map.cpp

Show First 20 Lines • Show All 97 Lines • ▼ Show 20 Lines#endif
return 0; return 0;
} }
// CHECK-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l32 // CHECK-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z3fooi_l32
// CHECK-SAME: (ptr noundef [[E:%.*]]) #[[ATTR0:[0-9]+]] { // CHECK-SAME: (ptr noundef [[E:%.*]]) #[[ATTR0:[0-9]+]] {
// CHECK-NEXT: entry: // CHECK-NEXT: entry:
// CHECK-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8 // CHECK-NEXT: [[E_ADDR:%.*]] = alloca ptr, align 8
// CHECK-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8 // CHECK-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [1 x ptr], align 8
// CHECK-NEXT: store ptr [[E]], ptr [[E_ADDR]], align 8 // CHECK-NEXT: store ptr [[E]], ptr [[E_ADDR]], align 8
// CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) // CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false)
// CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 // 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-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
// CHECK: user_code.entry: // CHECK: user_code.entry:
// CHECK-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]]) // CHECK-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]])
// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[E_ADDR]], align 8 // CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[E_ADDR]], align 8
// CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 // CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
// CHECK-NEXT: store ptr [[TMP2]], ptr [[TMP3]], align 8 // CHECK-NEXT: store ptr [[TMP2]], ptr [[TMP3]], align 8
// CHECK-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1) // CHECK-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB2]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 1)
// CHECK-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2, i1 true) // CHECK-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB1]], i8 2)
// CHECK-NEXT: ret void // CHECK-NEXT: ret void
// CHECK: worker.exit: // CHECK: worker.exit:
// CHECK-NEXT: ret void // CHECK-NEXT: ret void
// //
// //
// CHECK-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[E:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[E:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK-NEXT: entry: // CHECK-NEXT: entry:
▲ Show 20 Lines • Show All 1,946 Lines • Show Last 20 Lines

clang/test/OpenMP/remarks_parallel_in_multiple_target_state_machines.c

Show First 20 Lines • Show All 80 Lines • ▼ Show 20 Lines
#pragma omp target teams distribute parallel for #pragma omp target teams distribute parallel for
for (int i = 0; i < 100; ++i) { for (int i = 0; i < 100; ++i) {
} }
} }
#pragma omp begin declare target device_type(nohost) #pragma omp begin declare target device_type(nohost)
__attribute__((weak)) __attribute__((weak))
extern "C" int __kmpc_target_init(void *Ident, char Mode, extern "C" int __kmpc_target_init(void *Ident, char Mode,
bool UseGenericStateMachine, bool) { // all-remark {{Could not internalize function. Some optimizations may not be possible. [OMP140]}} bool UseGenericStateMachine) { // all-remark {{Could not internalize function. Some optimizations may not be possible. [OMP140]}}
return 0; return 0;
} }
#pragma omp end declare target #pragma omp end declare target
// all-remark@* 9 {{OpenMP runtime call __kmpc_global_thread_num deduplicated. [OMP170]}} // all-remark@* 9 {{OpenMP runtime call __kmpc_global_thread_num deduplicated. [OMP170]}}

clang/test/OpenMP/remarks_parallel_in_target_state_machine.c

Show All 36 Lines
#pragma omp target teams distribute parallel for #pragma omp target teams distribute parallel for
for (int i = 0; i < 100; ++i) { for (int i = 0; i < 100; ++i) {
} }
} }
#pragma omp begin declare target device_type(nohost) #pragma omp begin declare target device_type(nohost)
__attribute__((weak)) __attribute__((weak))
extern "C" int __kmpc_target_init(void *Ident, char Mode, extern "C" int __kmpc_target_init(void *Ident, char Mode,
bool UseGenericStateMachine, bool) { // expected-remark {{Could not internalize function. Some optimizations may not be possible. [OMP140]}} bool UseGenericStateMachine) { // expected-remark {{Could not internalize function. Some optimizations may not be possible. [OMP140]}}
return 0; return 0;
} }
#pragma omp end declare target #pragma omp end declare target
// expected-remark@* {{OpenMP runtime call __kmpc_global_thread_num deduplicated. [OMP170]}} // expected-remark@* {{OpenMP runtime call __kmpc_global_thread_num deduplicated. [OMP170]}}

clang/test/OpenMP/target_has_device_addr_codegen.cpp

Show First 20 Lines • Show All 344 Lines • ▼ Show 20 Lines
// CHECK-NEXT: br i1 [[TMP16]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK-NEXT: br i1 [[TMP16]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK: omp_offload.failed: // CHECK: omp_offload.failed:
// CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l145(ptr [[K]]) #[[ATTR5:[0-9]+]] // CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l145(ptr [[K]]) #[[ATTR5:[0-9]+]]
// CHECK-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK: omp_offload.cont: // CHECK: omp_offload.cont:
// CHECK-NEXT: [[TMP17:%.*]] = load ptr, ptr [[Z]], align 8 // CHECK-NEXT: [[TMP17:%.*]] = load ptr, ptr [[Z]], align 8
// CHECK-NEXT: store ptr [[TMP17]], ptr [[TMP]], align 8 // CHECK-NEXT: store ptr [[TMP17]], ptr [[TMP]], align 8
// CHECK-NEXT: [[TMP18:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK-NEXT: [[TMP18:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK-NEXT: [[TMP19:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK-NEXT: store ptr [[TMP18]], ptr [[TMP19]], align 8
// CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK-NEXT: store ptr [[TMP18]], ptr [[TMP20]], align 8 // CHECK-NEXT: store ptr [[TMP18]], ptr [[TMP20]], align 8
// CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK-NEXT: store ptr [[TMP18]], ptr [[TMP21]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP21]], align 8
// CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK-NEXT: store ptr null, ptr [[TMP22]], align 8 // CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK-NEXT: [[KERNEL_ARGS4:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK-NEXT: [[KERNEL_ARGS4:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 0 // CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 0
// CHECK-NEXT: store i32 1, ptr [[TMP24]], align 4
// CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 1
// CHECK-NEXT: store i32 1, ptr [[TMP25]], align 4 // CHECK-NEXT: store i32 1, ptr [[TMP25]], align 4
// CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 1 // CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 2
// CHECK-NEXT: store i32 1, ptr [[TMP26]], align 4 // CHECK-NEXT: store ptr [[TMP22]], ptr [[TMP26]], align 8
// CHECK-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 2 // CHECK-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 3
// CHECK-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 8 // CHECK-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 8
// CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 3 // CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 4
// CHECK-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 8 // CHECK-NEXT: store ptr @.offload_sizes.5, ptr [[TMP28]], align 8
// CHECK-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 4 // CHECK-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 5
// CHECK-NEXT: store ptr @.offload_sizes.5, ptr [[TMP29]], align 8 // CHECK-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP29]], align 8
// CHECK-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 5 // CHECK-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 6
// CHECK-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP30]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP30]], align 8
// CHECK-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 6 // CHECK-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 7
// CHECK-NEXT: store ptr null, ptr [[TMP31]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP31]], align 8
// CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 7 // CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 8
// CHECK-NEXT: store ptr null, ptr [[TMP32]], align 8 // CHECK-NEXT: store i64 0, ptr [[TMP32]], align 8
// CHECK-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 8 // CHECK-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l147.region_id, ptr [[KERNEL_ARGS4]])
// CHECK-NEXT: store i64 0, ptr [[TMP33]], align 8 // CHECK-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
// CHECK-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l147.region_id, ptr [[KERNEL_ARGS4]]) // CHECK-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
// CHECK-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK: omp_offload.failed5: // CHECK: omp_offload.failed5:
// CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l147(ptr [[TMP18]]) #[[ATTR5]] // CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l147(ptr [[TMP18]]) #[[ATTR5]]
// CHECK-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK: omp_offload.cont6: // CHECK: omp_offload.cont6:
// CHECK-NEXT: [[TMP36:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK-NEXT: store ptr [[AA]], ptr [[TMP35]], align 8
// CHECK-NEXT: [[TMP36:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK-NEXT: store ptr [[AA]], ptr [[TMP36]], align 8 // CHECK-NEXT: store ptr [[AA]], ptr [[TMP36]], align 8
// CHECK-NEXT: [[TMP37:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK-NEXT: [[TMP37:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0
// CHECK-NEXT: store ptr [[AA]], ptr [[TMP37]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP37]], align 8
// CHECK-NEXT: [[TMP38:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0 // CHECK-NEXT: [[TMP38:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK-NEXT: store ptr null, ptr [[TMP38]], align 8 // CHECK-NEXT: [[TMP39:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK-NEXT: [[TMP39:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK-NEXT: [[TMP40:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK-NEXT: [[KERNEL_ARGS10:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK-NEXT: [[KERNEL_ARGS10:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 0 // CHECK-NEXT: [[TMP40:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 0
// CHECK-NEXT: store i32 1, ptr [[TMP40]], align 4
// CHECK-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 1
// CHECK-NEXT: store i32 1, ptr [[TMP41]], align 4 // CHECK-NEXT: store i32 1, ptr [[TMP41]], align 4
// CHECK-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 1 // CHECK-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 2
// CHECK-NEXT: store i32 1, ptr [[TMP42]], align 4 // CHECK-NEXT: store ptr [[TMP38]], ptr [[TMP42]], align 8
// CHECK-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 2 // CHECK-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 3
// CHECK-NEXT: store ptr [[TMP39]], ptr [[TMP43]], align 8 // CHECK-NEXT: store ptr [[TMP39]], ptr [[TMP43]], align 8
// CHECK-NEXT: [[TMP44:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 3 // CHECK-NEXT: [[TMP44:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 4
// CHECK-NEXT: store ptr [[TMP40]], ptr [[TMP44]], align 8 // CHECK-NEXT: store ptr @.offload_sizes.7, ptr [[TMP44]], align 8
// CHECK-NEXT: [[TMP45:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 4 // CHECK-NEXT: [[TMP45:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 5
// CHECK-NEXT: store ptr @.offload_sizes.7, ptr [[TMP45]], align 8 // CHECK-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP45]], align 8
// CHECK-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 5 // CHECK-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 6
// CHECK-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP46]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP46]], align 8
// CHECK-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 6 // CHECK-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 7
// CHECK-NEXT: store ptr null, ptr [[TMP47]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP47]], align 8
// CHECK-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 7 // CHECK-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 8
// CHECK-NEXT: store ptr null, ptr [[TMP48]], align 8 // CHECK-NEXT: store i64 0, ptr [[TMP48]], align 8
// CHECK-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 8 // CHECK-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l149.region_id, ptr [[KERNEL_ARGS10]])
// CHECK-NEXT: store i64 0, ptr [[TMP49]], align 8 // CHECK-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0
// CHECK-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l149.region_id, ptr [[KERNEL_ARGS10]]) // CHECK-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0
// CHECK-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK: omp_offload.failed11: // CHECK: omp_offload.failed11:
// CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l149(ptr [[AA]]) #[[ATTR5]] // CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l149(ptr [[AA]]) #[[ATTR5]]
// CHECK-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK: omp_offload.cont12: // CHECK: omp_offload.cont12:
// CHECK-NEXT: [[TMP52:%.*]] = load ptr, ptr [[RAA]], align 8 // CHECK-NEXT: [[TMP51:%.*]] = load ptr, ptr [[RAA]], align 8
// CHECK-NEXT: store ptr [[TMP52]], ptr [[_TMP13]], align 8 // CHECK-NEXT: store ptr [[TMP51]], ptr [[_TMP13]], align 8
// CHECK-NEXT: [[TMP53:%.*]] = load ptr, ptr [[_TMP13]], align 8 // CHECK-NEXT: [[TMP52:%.*]] = load ptr, ptr [[_TMP13]], align 8
// CHECK-NEXT: [[TMP55:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0 // CHECK-NEXT: [[TMP53:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK-NEXT: store ptr [[TMP53]], ptr [[TMP55]], align 8 // CHECK-NEXT: store ptr [[TMP52]], ptr [[TMP53]], align 8
// CHECK-NEXT: [[TMP56:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS15]], i32 0, i32 0 // CHECK-NEXT: [[TMP54:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK-NEXT: store ptr [[TMP53]], ptr [[TMP56]], align 8 // CHECK-NEXT: store ptr [[TMP52]], ptr [[TMP54]], align 8
// CHECK-NEXT: [[TMP57:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS16]], i64 0, i64 0 // CHECK-NEXT: [[TMP55:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS16]], i64 0, i64 0
// CHECK-NEXT: store ptr null, ptr [[TMP57]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP55]], align 8
// CHECK-NEXT: [[TMP58:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0 // CHECK-NEXT: [[TMP56:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS14]], i32 0, i32 0
// CHECK-NEXT: [[TMP59:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS15]], i32 0, i32 0 // CHECK-NEXT: [[TMP57:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS15]], i32 0, i32 0
// CHECK-NEXT: [[KERNEL_ARGS17:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK-NEXT: [[KERNEL_ARGS17:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK-NEXT: [[TMP60:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 0 // CHECK-NEXT: [[TMP58:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 0
// CHECK-NEXT: store i32 1, ptr [[TMP60]], align 4 // CHECK-NEXT: store i32 1, ptr [[TMP58]], align 4
// CHECK-NEXT: [[TMP61:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 1 // CHECK-NEXT: [[TMP59:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 1
// CHECK-NEXT: store i32 1, ptr [[TMP61]], align 4 // CHECK-NEXT: store i32 1, ptr [[TMP59]], align 4
// CHECK-NEXT: [[TMP62:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 2 // CHECK-NEXT: [[TMP60:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 2
// CHECK-NEXT: store ptr [[TMP58]], ptr [[TMP62]], align 8 // CHECK-NEXT: store ptr [[TMP56]], ptr [[TMP60]], align 8
// CHECK-NEXT: [[TMP63:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 3 // CHECK-NEXT: [[TMP61:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 3
// CHECK-NEXT: store ptr [[TMP59]], ptr [[TMP63]], align 8 // CHECK-NEXT: store ptr [[TMP57]], ptr [[TMP61]], align 8
// CHECK-NEXT: [[TMP64:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 4 // CHECK-NEXT: [[TMP62:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 4
// CHECK-NEXT: store ptr @.offload_sizes.9, ptr [[TMP64]], align 8 // CHECK-NEXT: store ptr @.offload_sizes.9, ptr [[TMP62]], align 8
// CHECK-NEXT: [[TMP65:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 5 // CHECK-NEXT: [[TMP63:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 5
// CHECK-NEXT: store ptr @.offload_maptypes.10, ptr [[TMP65]], align 8 // CHECK-NEXT: store ptr @.offload_maptypes.10, ptr [[TMP63]], align 8
// CHECK-NEXT: [[TMP66:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 6 // CHECK-NEXT: [[TMP64:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 6
// CHECK-NEXT: store ptr null, ptr [[TMP66]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP64]], align 8
// CHECK-NEXT: [[TMP67:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 7 // CHECK-NEXT: [[TMP65:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 7
// CHECK-NEXT: store ptr null, ptr [[TMP67]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP65]], align 8
// CHECK-NEXT: [[TMP68:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 8 // CHECK-NEXT: [[TMP66:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS17]], i32 0, i32 8
// CHECK-NEXT: store i64 0, ptr [[TMP68]], align 8 // CHECK-NEXT: store i64 0, ptr [[TMP66]], align 8
// CHECK-NEXT: [[TMP69:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l151.region_id, ptr [[KERNEL_ARGS17]]) // CHECK-NEXT: [[TMP67:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l151.region_id, ptr [[KERNEL_ARGS17]])
// CHECK-NEXT: [[TMP70:%.*]] = icmp ne i32 [[TMP69]], 0 // CHECK-NEXT: [[TMP68:%.*]] = icmp ne i32 [[TMP67]], 0
// CHECK-NEXT: br i1 [[TMP70]], label [[OMP_OFFLOAD_FAILED18:%.*]], label [[OMP_OFFLOAD_CONT19:%.*]] // CHECK-NEXT: br i1 [[TMP68]], label [[OMP_OFFLOAD_FAILED18:%.*]], label [[OMP_OFFLOAD_CONT19:%.*]]
// CHECK: omp_offload.failed18: // CHECK: omp_offload.failed18:
// CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l151(ptr [[TMP53]]) #[[ATTR5]] // CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l151(ptr [[TMP52]]) #[[ATTR5]]
// CHECK-NEXT: br label [[OMP_OFFLOAD_CONT19]] // CHECK-NEXT: br label [[OMP_OFFLOAD_CONT19]]
// CHECK: omp_offload.cont19: // CHECK: omp_offload.cont19:
// CHECK-NEXT: [[TMP71:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK-NEXT: [[TMP69:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK-NEXT: store ptr [[H]], ptr [[TMP71]], align 8 // CHECK-NEXT: store ptr [[H]], ptr [[TMP69]], align 8
// CHECK-NEXT: [[TMP72:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK-NEXT: [[TMP70:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK-NEXT: store ptr [[H]], ptr [[TMP72]], align 8 // CHECK-NEXT: store ptr [[H]], ptr [[TMP70]], align 8
// CHECK-NEXT: [[TMP73:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0 // CHECK-NEXT: [[TMP71:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0
// CHECK-NEXT: store ptr null, ptr [[TMP73]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP71]], align 8
// CHECK-NEXT: [[TMP74:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 // CHECK-NEXT: [[TMP72:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0
// CHECK-NEXT: [[TMP75:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 // CHECK-NEXT: [[TMP73:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS21]], i32 0, i32 0
// CHECK-NEXT: [[KERNEL_ARGS23:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK-NEXT: [[KERNEL_ARGS23:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK-NEXT: [[TMP76:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 0 // CHECK-NEXT: [[TMP74:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 0
// CHECK-NEXT: store i32 1, ptr [[TMP76]], align 4 // CHECK-NEXT: store i32 1, ptr [[TMP74]], align 4
// CHECK-NEXT: [[TMP77:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 1 // CHECK-NEXT: [[TMP75:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 1
// CHECK-NEXT: store i32 1, ptr [[TMP77]], align 4 // CHECK-NEXT: store i32 1, ptr [[TMP75]], align 4
// CHECK-NEXT: [[TMP78:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 2 // CHECK-NEXT: [[TMP76:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 2
// CHECK-NEXT: store ptr [[TMP74]], ptr [[TMP78]], align 8 // CHECK-NEXT: store ptr [[TMP72]], ptr [[TMP76]], align 8
// CHECK-NEXT: [[TMP79:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 3 // CHECK-NEXT: [[TMP77:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 3
// CHECK-NEXT: store ptr [[TMP75]], ptr [[TMP79]], align 8 // CHECK-NEXT: store ptr [[TMP73]], ptr [[TMP77]], align 8
// CHECK-NEXT: [[TMP80:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 4 // CHECK-NEXT: [[TMP78:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 4
// CHECK-NEXT: store ptr @.offload_sizes.11, ptr [[TMP80]], align 8 // CHECK-NEXT: store ptr @.offload_sizes.11, ptr [[TMP78]], align 8
// CHECK-NEXT: [[TMP81:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 5 // CHECK-NEXT: [[TMP79:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 5
// CHECK-NEXT: store ptr @.offload_maptypes.12, ptr [[TMP81]], align 8 // CHECK-NEXT: store ptr @.offload_maptypes.12, ptr [[TMP79]], align 8
// CHECK-NEXT: [[TMP82:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 6 // CHECK-NEXT: [[TMP80:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 6
// CHECK-NEXT: store ptr null, ptr [[TMP82]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP80]], align 8
// CHECK-NEXT: [[TMP83:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 7 // CHECK-NEXT: [[TMP81:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 7
// CHECK-NEXT: store ptr null, ptr [[TMP83]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP81]], align 8
// CHECK-NEXT: [[TMP84:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 8 // CHECK-NEXT: [[TMP82:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS23]], i32 0, i32 8
// CHECK-NEXT: store i64 0, ptr [[TMP84]], align 8 // CHECK-NEXT: store i64 0, ptr [[TMP82]], align 8
// CHECK-NEXT: [[TMP85:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.region_id, ptr [[KERNEL_ARGS23]]) // CHECK-NEXT: [[TMP83:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153.region_id, ptr [[KERNEL_ARGS23]])
// CHECK-NEXT: [[TMP86:%.*]] = icmp ne i32 [[TMP85]], 0 // CHECK-NEXT: [[TMP84:%.*]] = icmp ne i32 [[TMP83]], 0
// CHECK-NEXT: br i1 [[TMP86]], label [[OMP_OFFLOAD_FAILED24:%.*]], label [[OMP_OFFLOAD_CONT25:%.*]] // CHECK-NEXT: br i1 [[TMP84]], label [[OMP_OFFLOAD_FAILED24:%.*]], label [[OMP_OFFLOAD_CONT25:%.*]]
// CHECK: omp_offload.failed24: // CHECK: omp_offload.failed24:
// CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153(ptr [[H]]) #[[ATTR5]] // CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l153(ptr [[H]]) #[[ATTR5]]
// CHECK-NEXT: br label [[OMP_OFFLOAD_CONT25]] // CHECK-NEXT: br label [[OMP_OFFLOAD_CONT25]]
// CHECK: omp_offload.cont25: // CHECK: omp_offload.cont25:
// CHECK-NEXT: [[TMP87:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0 // CHECK-NEXT: [[TMP85:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0
// CHECK-NEXT: store ptr [[DA]], ptr [[TMP87]], align 8 // CHECK-NEXT: store ptr [[DA]], ptr [[TMP85]], align 8
// CHECK-NEXT: [[TMP88:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS27]], i32 0, i32 0 // CHECK-NEXT: [[TMP86:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS27]], i32 0, i32 0
// CHECK-NEXT: store ptr [[DA]], ptr [[TMP88]], align 8 // CHECK-NEXT: store ptr [[DA]], ptr [[TMP86]], align 8
// CHECK-NEXT: [[TMP89:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 0 // CHECK-NEXT: [[TMP87:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS28]], i64 0, i64 0
// CHECK-NEXT: store ptr null, ptr [[TMP89]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP87]], align 8
// CHECK-NEXT: [[TMP90:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0 // CHECK-NEXT: [[TMP88:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS26]], i32 0, i32 0
// CHECK-NEXT: [[TMP91:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS27]], i32 0, i32 0 // CHECK-NEXT: [[TMP89:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS27]], i32 0, i32 0
// CHECK-NEXT: [[KERNEL_ARGS29:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK-NEXT: [[KERNEL_ARGS29:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK-NEXT: [[TMP92:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 0 // CHECK-NEXT: [[TMP90:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 0
// CHECK-NEXT: store i32 1, ptr [[TMP92]], align 4 // CHECK-NEXT: store i32 1, ptr [[TMP90]], align 4
// CHECK-NEXT: [[TMP93:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 1 // CHECK-NEXT: [[TMP91:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 1
// CHECK-NEXT: store i32 1, ptr [[TMP93]], align 4 // CHECK-NEXT: store i32 1, ptr [[TMP91]], align 4
// CHECK-NEXT: [[TMP94:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 2 // CHECK-NEXT: [[TMP92:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 2
// CHECK-NEXT: store ptr [[TMP90]], ptr [[TMP94]], align 8 // CHECK-NEXT: store ptr [[TMP88]], ptr [[TMP92]], align 8
// CHECK-NEXT: [[TMP95:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 3 // CHECK-NEXT: [[TMP93:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 3
// CHECK-NEXT: store ptr [[TMP91]], ptr [[TMP95]], align 8 // CHECK-NEXT: store ptr [[TMP89]], ptr [[TMP93]], align 8
// CHECK-NEXT: [[TMP96:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 4 // CHECK-NEXT: [[TMP94:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 4
// CHECK-NEXT: store ptr @.offload_sizes.13, ptr [[TMP96]], align 8 // CHECK-NEXT: store ptr @.offload_sizes.13, ptr [[TMP94]], align 8
// CHECK-NEXT: [[TMP97:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 5 // CHECK-NEXT: [[TMP95:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 5
// CHECK-NEXT: store ptr @.offload_maptypes.14, ptr [[TMP97]], align 8 // CHECK-NEXT: store ptr @.offload_maptypes.14, ptr [[TMP95]], align 8
// CHECK-NEXT: [[TMP98:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 6 // CHECK-NEXT: [[TMP96:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 6
// CHECK-NEXT: store ptr null, ptr [[TMP98]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP96]], align 8
// CHECK-NEXT: [[TMP99:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 7 // CHECK-NEXT: [[TMP97:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 7
// CHECK-NEXT: store ptr null, ptr [[TMP99]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP97]], align 8
// CHECK-NEXT: [[TMP100:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 8 // CHECK-NEXT: [[TMP98:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS29]], i32 0, i32 8
// CHECK-NEXT: store i64 0, ptr [[TMP100]], align 8 // CHECK-NEXT: store i64 0, ptr [[TMP98]], align 8
// CHECK-NEXT: [[TMP101:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l155.region_id, ptr [[KERNEL_ARGS29]]) // CHECK-NEXT: [[TMP99:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l155.region_id, ptr [[KERNEL_ARGS29]])
// CHECK-NEXT: [[TMP102:%.*]] = icmp ne i32 [[TMP101]], 0 // CHECK-NEXT: [[TMP100:%.*]] = icmp ne i32 [[TMP99]], 0
// CHECK-NEXT: br i1 [[TMP102]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]] // CHECK-NEXT: br i1 [[TMP100]], label [[OMP_OFFLOAD_FAILED30:%.*]], label [[OMP_OFFLOAD_CONT31:%.*]]
// CHECK: omp_offload.failed30: // CHECK: omp_offload.failed30:
// CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l155(ptr [[DA]]) #[[ATTR5]] // CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l155(ptr [[DA]]) #[[ATTR5]]
// CHECK-NEXT: br label [[OMP_OFFLOAD_CONT31]] // CHECK-NEXT: br label [[OMP_OFFLOAD_CONT31]]
// CHECK: omp_offload.cont31: // CHECK: omp_offload.cont31:
// CHECK-NEXT: [[TMP103:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK-NEXT: [[TMP101:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_S0_(i32 noundef signext [[TMP103]]) // CHECK-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_S0_(i32 noundef signext [[TMP101]])
// CHECK-NEXT: [[CALL32:%.*]] = call noundef ptr @_Z5tmainIPiET_S1_(ptr noundef [[ARGC_ADDR]]) // CHECK-NEXT: [[CALL32:%.*]] = call noundef ptr @_Z5tmainIPiET_S1_(ptr noundef [[ARGC_ADDR]])
// CHECK-NEXT: [[TMP104:%.*]] = load i32, ptr [[CALL32]], align 4 // CHECK-NEXT: [[TMP102:%.*]] = load i32, ptr [[CALL32]], align 4
// CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[CALL]], [[TMP104]] // CHECK-NEXT: [[ADD:%.*]] = add nsw i32 [[CALL]], [[TMP102]]
// CHECK-NEXT: ret i32 [[ADD]] // CHECK-NEXT: ret i32 [[ADD]]
// //
// //
// CHECK-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l145 // CHECK-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l145
// CHECK-SAME: (ptr noundef nonnull align 8 dereferenceable(8) [[K:%.*]]) #[[ATTR4:[0-9]+]] { // CHECK-SAME: (ptr noundef nonnull align 8 dereferenceable(8) [[K:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK-NEXT: entry: // CHECK-NEXT: entry:
// CHECK-NEXT: [[K_ADDR:%.*]] = alloca ptr, align 8 // CHECK-NEXT: [[K_ADDR:%.*]] = alloca ptr, align 8
// CHECK-NEXT: store ptr [[K]], ptr [[K_ADDR]], align 8 // CHECK-NEXT: store ptr [[K]], ptr [[K_ADDR]], align 8
▲ Show 20 Lines • Show All 137 Lines • ▼ Show 20 Lines
// CHECK-NEXT: br i1 [[TMP16]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK-NEXT: br i1 [[TMP16]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK: omp_offload.failed: // CHECK: omp_offload.failed:
// CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_S0__l123(ptr [[K]]) #[[ATTR5]] // CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_S0__l123(ptr [[K]]) #[[ATTR5]]
// CHECK-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK: omp_offload.cont: // CHECK: omp_offload.cont:
// CHECK-NEXT: [[TMP17:%.*]] = load ptr, ptr [[Z]], align 8 // CHECK-NEXT: [[TMP17:%.*]] = load ptr, ptr [[Z]], align 8
// CHECK-NEXT: store ptr [[TMP17]], ptr [[TMP]], align 8 // CHECK-NEXT: store ptr [[TMP17]], ptr [[TMP]], align 8
// CHECK-NEXT: [[TMP18:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK-NEXT: [[TMP18:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK-NEXT: [[TMP19:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK-NEXT: store ptr [[TMP18]], ptr [[TMP19]], align 8
// CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK-NEXT: store ptr [[TMP18]], ptr [[TMP20]], align 8 // CHECK-NEXT: store ptr [[TMP18]], ptr [[TMP20]], align 8
// CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK-NEXT: store ptr [[TMP18]], ptr [[TMP21]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP21]], align 8
// CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK-NEXT: store ptr null, ptr [[TMP22]], align 8 // CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK-NEXT: [[KERNEL_ARGS4:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK-NEXT: [[KERNEL_ARGS4:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 0 // CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 0
// CHECK-NEXT: store i32 1, ptr [[TMP24]], align 4
// CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 1
// CHECK-NEXT: store i32 1, ptr [[TMP25]], align 4 // CHECK-NEXT: store i32 1, ptr [[TMP25]], align 4
// CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 1 // CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 2
// CHECK-NEXT: store i32 1, ptr [[TMP26]], align 4 // CHECK-NEXT: store ptr [[TMP22]], ptr [[TMP26]], align 8
// CHECK-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 2 // CHECK-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 3
// CHECK-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 8 // CHECK-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 8
// CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 3 // CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 4
// CHECK-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 8 // CHECK-NEXT: store ptr @.offload_sizes.17, ptr [[TMP28]], align 8
// CHECK-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 4 // CHECK-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 5
// CHECK-NEXT: store ptr @.offload_sizes.17, ptr [[TMP29]], align 8 // CHECK-NEXT: store ptr @.offload_maptypes.18, ptr [[TMP29]], align 8
// CHECK-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 5 // CHECK-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 6
// CHECK-NEXT: store ptr @.offload_maptypes.18, ptr [[TMP30]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP30]], align 8
// CHECK-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 6 // CHECK-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 7
// CHECK-NEXT: store ptr null, ptr [[TMP31]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP31]], align 8
// CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 7 // CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 8
// CHECK-NEXT: store ptr null, ptr [[TMP32]], align 8 // CHECK-NEXT: store i64 0, ptr [[TMP32]], align 8
// CHECK-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 8 // CHECK-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_S0__l125.region_id, ptr [[KERNEL_ARGS4]])
// CHECK-NEXT: store i64 0, ptr [[TMP33]], align 8 // CHECK-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
// CHECK-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_S0__l125.region_id, ptr [[KERNEL_ARGS4]]) // CHECK-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
// CHECK-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK: omp_offload.failed5: // CHECK: omp_offload.failed5:
// CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_S0__l125(ptr [[TMP18]]) #[[ATTR5]] // CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_S0__l125(ptr [[TMP18]]) #[[ATTR5]]
// CHECK-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK: omp_offload.cont6: // CHECK: omp_offload.cont6:
// CHECK-NEXT: [[TMP36:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK-NEXT: store ptr [[AA]], ptr [[TMP35]], align 8
// CHECK-NEXT: [[TMP36:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK-NEXT: store ptr [[AA]], ptr [[TMP36]], align 8 // CHECK-NEXT: store ptr [[AA]], ptr [[TMP36]], align 8
// CHECK-NEXT: [[TMP37:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK-NEXT: [[TMP37:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0
// CHECK-NEXT: store ptr [[AA]], ptr [[TMP37]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP37]], align 8
// CHECK-NEXT: [[TMP38:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0 // CHECK-NEXT: [[TMP38:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK-NEXT: store ptr null, ptr [[TMP38]], align 8 // CHECK-NEXT: [[TMP39:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK-NEXT: [[TMP39:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK-NEXT: [[TMP40:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK-NEXT: [[KERNEL_ARGS10:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK-NEXT: [[KERNEL_ARGS10:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 0 // CHECK-NEXT: [[TMP40:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 0
// CHECK-NEXT: store i32 1, ptr [[TMP40]], align 4
// CHECK-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 1
// CHECK-NEXT: store i32 1, ptr [[TMP41]], align 4 // CHECK-NEXT: store i32 1, ptr [[TMP41]], align 4
// CHECK-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 1 // CHECK-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 2
// CHECK-NEXT: store i32 1, ptr [[TMP42]], align 4 // CHECK-NEXT: store ptr [[TMP38]], ptr [[TMP42]], align 8
// CHECK-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 2 // CHECK-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 3
// CHECK-NEXT: store ptr [[TMP39]], ptr [[TMP43]], align 8 // CHECK-NEXT: store ptr [[TMP39]], ptr [[TMP43]], align 8
// CHECK-NEXT: [[TMP44:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 3 // CHECK-NEXT: [[TMP44:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 4
// CHECK-NEXT: store ptr [[TMP40]], ptr [[TMP44]], align 8 // CHECK-NEXT: store ptr @.offload_sizes.19, ptr [[TMP44]], align 8
// CHECK-NEXT: [[TMP45:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 4 // CHECK-NEXT: [[TMP45:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 5
// CHECK-NEXT: store ptr @.offload_sizes.19, ptr [[TMP45]], align 8 // CHECK-NEXT: store ptr @.offload_maptypes.20, ptr [[TMP45]], align 8
// CHECK-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 5 // CHECK-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 6
// CHECK-NEXT: store ptr @.offload_maptypes.20, ptr [[TMP46]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP46]], align 8
// CHECK-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 6 // CHECK-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 7
// CHECK-NEXT: store ptr null, ptr [[TMP47]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP47]], align 8
// CHECK-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 7 // CHECK-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 8
// CHECK-NEXT: store ptr null, ptr [[TMP48]], align 8 // CHECK-NEXT: store i64 0, ptr [[TMP48]], align 8
// CHECK-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 8 // CHECK-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_S0__l127.region_id, ptr [[KERNEL_ARGS10]])
// CHECK-NEXT: store i64 0, ptr [[TMP49]], align 8 // CHECK-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0
// CHECK-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_S0__l127.region_id, ptr [[KERNEL_ARGS10]]) // CHECK-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0
// CHECK-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK: omp_offload.failed11: // CHECK: omp_offload.failed11:
// CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_S0__l127(ptr [[AA]]) #[[ATTR5]] // CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_S0__l127(ptr [[AA]]) #[[ATTR5]]
// CHECK-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK: omp_offload.cont12: // CHECK: omp_offload.cont12:
// CHECK-NEXT: [[TMP52:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0 // CHECK-NEXT: [[TMP51:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK-NEXT: store ptr [[H]], ptr [[TMP51]], align 8
// CHECK-NEXT: [[TMP52:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK-NEXT: store ptr [[H]], ptr [[TMP52]], align 8 // CHECK-NEXT: store ptr [[H]], ptr [[TMP52]], align 8
// CHECK-NEXT: [[TMP53:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS14]], i32 0, i32 0 // CHECK-NEXT: [[TMP53:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS15]], i64 0, i64 0
// CHECK-NEXT: store ptr [[H]], ptr [[TMP53]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP53]], align 8
// CHECK-NEXT: [[TMP54:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS15]], i64 0, i64 0 // CHECK-NEXT: [[TMP54:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK-NEXT: store ptr null, ptr [[TMP54]], align 8 // CHECK-NEXT: [[TMP55:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK-NEXT: [[TMP55:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK-NEXT: [[TMP56:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK-NEXT: [[KERNEL_ARGS16:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK-NEXT: [[KERNEL_ARGS16:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK-NEXT: [[TMP57:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 0 // CHECK-NEXT: [[TMP56:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 0
// CHECK-NEXT: store i32 1, ptr [[TMP56]], align 4
// CHECK-NEXT: [[TMP57:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 1
// CHECK-NEXT: store i32 1, ptr [[TMP57]], align 4 // CHECK-NEXT: store i32 1, ptr [[TMP57]], align 4
// CHECK-NEXT: [[TMP58:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 1 // CHECK-NEXT: [[TMP58:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 2
// CHECK-NEXT: store i32 1, ptr [[TMP58]], align 4 // CHECK-NEXT: store ptr [[TMP54]], ptr [[TMP58]], align 8
// CHECK-NEXT: [[TMP59:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 2 // CHECK-NEXT: [[TMP59:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 3
// CHECK-NEXT: store ptr [[TMP55]], ptr [[TMP59]], align 8 // CHECK-NEXT: store ptr [[TMP55]], ptr [[TMP59]], align 8
// CHECK-NEXT: [[TMP60:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 3 // CHECK-NEXT: [[TMP60:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 4
// CHECK-NEXT: store ptr [[TMP56]], ptr [[TMP60]], align 8 // CHECK-NEXT: store ptr @.offload_sizes.21, ptr [[TMP60]], align 8
// CHECK-NEXT: [[TMP61:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 4 // CHECK-NEXT: [[TMP61:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 5
// CHECK-NEXT: store ptr @.offload_sizes.21, ptr [[TMP61]], align 8 // CHECK-NEXT: store ptr @.offload_maptypes.22, ptr [[TMP61]], align 8
// CHECK-NEXT: [[TMP62:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 5 // CHECK-NEXT: [[TMP62:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 6
// CHECK-NEXT: store ptr @.offload_maptypes.22, ptr [[TMP62]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP62]], align 8
// CHECK-NEXT: [[TMP63:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 6 // CHECK-NEXT: [[TMP63:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 7
// CHECK-NEXT: store ptr null, ptr [[TMP63]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP63]], align 8
// CHECK-NEXT: [[TMP64:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 7 // CHECK-NEXT: [[TMP64:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 8
// CHECK-NEXT: store ptr null, ptr [[TMP64]], align 8 // CHECK-NEXT: store i64 0, ptr [[TMP64]], align 8
// CHECK-NEXT: [[TMP65:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 8 // CHECK-NEXT: [[TMP65:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_S0__l129.region_id, ptr [[KERNEL_ARGS16]])
// CHECK-NEXT: store i64 0, ptr [[TMP65]], align 8 // CHECK-NEXT: [[TMP66:%.*]] = icmp ne i32 [[TMP65]], 0
// CHECK-NEXT: [[TMP66:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_S0__l129.region_id, ptr [[KERNEL_ARGS16]]) // CHECK-NEXT: br i1 [[TMP66]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK-NEXT: [[TMP67:%.*]] = icmp ne i32 [[TMP66]], 0
// CHECK-NEXT: br i1 [[TMP67]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK: omp_offload.failed17: // CHECK: omp_offload.failed17:
// CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_S0__l129(ptr [[H]]) #[[ATTR5]] // CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_S0__l129(ptr [[H]]) #[[ATTR5]]
// CHECK-NEXT: br label [[OMP_OFFLOAD_CONT18]] // CHECK-NEXT: br label [[OMP_OFFLOAD_CONT18]]
// CHECK: omp_offload.cont18: // CHECK: omp_offload.cont18:
// CHECK-NEXT: ret i32 0 // CHECK-NEXT: ret i32 0
// //
// //
// CHECK-LABEL: define {{[^@]+}}@_Z5tmainIPiET_S1_ // CHECK-LABEL: define {{[^@]+}}@_Z5tmainIPiET_S1_
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
// CHECK-NEXT: br i1 [[TMP16]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK-NEXT: br i1 [[TMP16]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]
// CHECK: omp_offload.failed: // CHECK: omp_offload.failed:
// CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPiET_S1__l123(ptr [[K]]) #[[ATTR5]] // CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPiET_S1__l123(ptr [[K]]) #[[ATTR5]]
// CHECK-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK: omp_offload.cont: // CHECK: omp_offload.cont:
// CHECK-NEXT: [[TMP17:%.*]] = load ptr, ptr [[Z]], align 8 // CHECK-NEXT: [[TMP17:%.*]] = load ptr, ptr [[Z]], align 8
// CHECK-NEXT: store ptr [[TMP17]], ptr [[TMP]], align 8 // CHECK-NEXT: store ptr [[TMP17]], ptr [[TMP]], align 8
// CHECK-NEXT: [[TMP18:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK-NEXT: [[TMP18:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK-NEXT: [[TMP19:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK-NEXT: store ptr [[TMP18]], ptr [[TMP19]], align 8
// CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK-NEXT: store ptr [[TMP18]], ptr [[TMP20]], align 8 // CHECK-NEXT: store ptr [[TMP18]], ptr [[TMP20]], align 8
// CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK-NEXT: store ptr [[TMP18]], ptr [[TMP21]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP21]], align 8
// CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK-NEXT: store ptr null, ptr [[TMP22]], align 8 // CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK-NEXT: [[KERNEL_ARGS4:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK-NEXT: [[KERNEL_ARGS4:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 0 // CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 0
// CHECK-NEXT: store i32 1, ptr [[TMP24]], align 4
// CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 1
// CHECK-NEXT: store i32 1, ptr [[TMP25]], align 4 // CHECK-NEXT: store i32 1, ptr [[TMP25]], align 4
// CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 1 // CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 2
// CHECK-NEXT: store i32 1, ptr [[TMP26]], align 4 // CHECK-NEXT: store ptr [[TMP22]], ptr [[TMP26]], align 8
// CHECK-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 2 // CHECK-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 3
// CHECK-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 8 // CHECK-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 8
// CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 3 // CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 4
// CHECK-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 8 // CHECK-NEXT: store ptr @.offload_sizes.25, ptr [[TMP28]], align 8
// CHECK-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 4 // CHECK-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 5
// CHECK-NEXT: store ptr @.offload_sizes.25, ptr [[TMP29]], align 8 // CHECK-NEXT: store ptr @.offload_maptypes.26, ptr [[TMP29]], align 8
// CHECK-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 5 // CHECK-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 6
// CHECK-NEXT: store ptr @.offload_maptypes.26, ptr [[TMP30]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP30]], align 8
// CHECK-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 6 // CHECK-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 7
// CHECK-NEXT: store ptr null, ptr [[TMP31]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP31]], align 8
// CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 7 // CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 8
// CHECK-NEXT: store ptr null, ptr [[TMP32]], align 8 // CHECK-NEXT: store i64 0, ptr [[TMP32]], align 8
// CHECK-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 8 // CHECK-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPiET_S1__l125.region_id, ptr [[KERNEL_ARGS4]])
// CHECK-NEXT: store i64 0, ptr [[TMP33]], align 8 // CHECK-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
// CHECK-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPiET_S1__l125.region_id, ptr [[KERNEL_ARGS4]]) // CHECK-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0
// CHECK-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK: omp_offload.failed5: // CHECK: omp_offload.failed5:
// CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPiET_S1__l125(ptr [[TMP18]]) #[[ATTR5]] // CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPiET_S1__l125(ptr [[TMP18]]) #[[ATTR5]]
// CHECK-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK: omp_offload.cont6: // CHECK: omp_offload.cont6:
// CHECK-NEXT: [[TMP36:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0 // CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK-NEXT: store ptr [[AA]], ptr [[TMP35]], align 8
// CHECK-NEXT: [[TMP36:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK-NEXT: store ptr [[AA]], ptr [[TMP36]], align 8 // CHECK-NEXT: store ptr [[AA]], ptr [[TMP36]], align 8
// CHECK-NEXT: [[TMP37:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 0 // CHECK-NEXT: [[TMP37:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0
// CHECK-NEXT: store ptr [[AA]], ptr [[TMP37]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP37]], align 8
// CHECK-NEXT: [[TMP38:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS9]], i64 0, i64 0 // CHECK-NEXT: [[TMP38:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK-NEXT: store ptr null, ptr [[TMP38]], align 8 // CHECK-NEXT: [[TMP39:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK-NEXT: [[TMP39:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS7]], i32 0, i32 0
// CHECK-NEXT: [[TMP40:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS8]], i32 0, i32 0
// CHECK-NEXT: [[KERNEL_ARGS10:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK-NEXT: [[KERNEL_ARGS10:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 0 // CHECK-NEXT: [[TMP40:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 0
// CHECK-NEXT: store i32 1, ptr [[TMP40]], align 4
// CHECK-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 1
// CHECK-NEXT: store i32 1, ptr [[TMP41]], align 4 // CHECK-NEXT: store i32 1, ptr [[TMP41]], align 4
// CHECK-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 1 // CHECK-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 2
// CHECK-NEXT: store i32 1, ptr [[TMP42]], align 4 // CHECK-NEXT: store ptr [[TMP38]], ptr [[TMP42]], align 8
// CHECK-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 2 // CHECK-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 3
// CHECK-NEXT: store ptr [[TMP39]], ptr [[TMP43]], align 8 // CHECK-NEXT: store ptr [[TMP39]], ptr [[TMP43]], align 8
// CHECK-NEXT: [[TMP44:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 3 // CHECK-NEXT: [[TMP44:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 4
// CHECK-NEXT: store ptr [[TMP40]], ptr [[TMP44]], align 8 // CHECK-NEXT: store ptr @.offload_sizes.27, ptr [[TMP44]], align 8
// CHECK-NEXT: [[TMP45:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 4 // CHECK-NEXT: [[TMP45:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 5
// CHECK-NEXT: store ptr @.offload_sizes.27, ptr [[TMP45]], align 8 // CHECK-NEXT: store ptr @.offload_maptypes.28, ptr [[TMP45]], align 8
// CHECK-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 5 // CHECK-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 6
// CHECK-NEXT: store ptr @.offload_maptypes.28, ptr [[TMP46]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP46]], align 8
// CHECK-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 6 // CHECK-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 7
// CHECK-NEXT: store ptr null, ptr [[TMP47]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP47]], align 8
// CHECK-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 7 // CHECK-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 8
// CHECK-NEXT: store ptr null, ptr [[TMP48]], align 8 // CHECK-NEXT: store i64 0, ptr [[TMP48]], align 8
// CHECK-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS10]], i32 0, i32 8 // CHECK-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPiET_S1__l127.region_id, ptr [[KERNEL_ARGS10]])
// CHECK-NEXT: store i64 0, ptr [[TMP49]], align 8 // CHECK-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0
// CHECK-NEXT: [[TMP50:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPiET_S1__l127.region_id, ptr [[KERNEL_ARGS10]]) // CHECK-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK-NEXT: [[TMP51:%.*]] = icmp ne i32 [[TMP50]], 0
// CHECK-NEXT: br i1 [[TMP51]], label [[OMP_OFFLOAD_FAILED11:%.*]], label [[OMP_OFFLOAD_CONT12:%.*]]
// CHECK: omp_offload.failed11: // CHECK: omp_offload.failed11:
// CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPiET_S1__l127(ptr [[AA]]) #[[ATTR5]] // CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPiET_S1__l127(ptr [[AA]]) #[[ATTR5]]
// CHECK-NEXT: br label [[OMP_OFFLOAD_CONT12]] // CHECK-NEXT: br label [[OMP_OFFLOAD_CONT12]]
// CHECK: omp_offload.cont12: // CHECK: omp_offload.cont12:
// CHECK-NEXT: [[TMP52:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0 // CHECK-NEXT: [[TMP51:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK-NEXT: store ptr [[H]], ptr [[TMP51]], align 8
// CHECK-NEXT: [[TMP52:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK-NEXT: store ptr [[H]], ptr [[TMP52]], align 8 // CHECK-NEXT: store ptr [[H]], ptr [[TMP52]], align 8
// CHECK-NEXT: [[TMP53:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS14]], i32 0, i32 0 // CHECK-NEXT: [[TMP53:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS15]], i64 0, i64 0
// CHECK-NEXT: store ptr [[H]], ptr [[TMP53]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP53]], align 8
// CHECK-NEXT: [[TMP54:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS15]], i64 0, i64 0 // CHECK-NEXT: [[TMP54:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK-NEXT: store ptr null, ptr [[TMP54]], align 8 // CHECK-NEXT: [[TMP55:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK-NEXT: [[TMP55:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS13]], i32 0, i32 0
// CHECK-NEXT: [[TMP56:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS14]], i32 0, i32 0
// CHECK-NEXT: [[KERNEL_ARGS16:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK-NEXT: [[KERNEL_ARGS16:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK-NEXT: [[TMP57:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 0 // CHECK-NEXT: [[TMP56:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 0
// CHECK-NEXT: store i32 1, ptr [[TMP56]], align 4
// CHECK-NEXT: [[TMP57:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 1
// CHECK-NEXT: store i32 1, ptr [[TMP57]], align 4 // CHECK-NEXT: store i32 1, ptr [[TMP57]], align 4
// CHECK-NEXT: [[TMP58:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 1 // CHECK-NEXT: [[TMP58:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 2
// CHECK-NEXT: store i32 1, ptr [[TMP58]], align 4 // CHECK-NEXT: store ptr [[TMP54]], ptr [[TMP58]], align 8
// CHECK-NEXT: [[TMP59:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 2 // CHECK-NEXT: [[TMP59:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 3
// CHECK-NEXT: store ptr [[TMP55]], ptr [[TMP59]], align 8 // CHECK-NEXT: store ptr [[TMP55]], ptr [[TMP59]], align 8
// CHECK-NEXT: [[TMP60:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 3 // CHECK-NEXT: [[TMP60:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 4
// CHECK-NEXT: store ptr [[TMP56]], ptr [[TMP60]], align 8 // CHECK-NEXT: store ptr @.offload_sizes.29, ptr [[TMP60]], align 8
// CHECK-NEXT: [[TMP61:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 4 // CHECK-NEXT: [[TMP61:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 5
// CHECK-NEXT: store ptr @.offload_sizes.29, ptr [[TMP61]], align 8 // CHECK-NEXT: store ptr @.offload_maptypes.30, ptr [[TMP61]], align 8
// CHECK-NEXT: [[TMP62:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 5 // CHECK-NEXT: [[TMP62:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 6
// CHECK-NEXT: store ptr @.offload_maptypes.30, ptr [[TMP62]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP62]], align 8
// CHECK-NEXT: [[TMP63:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 6 // CHECK-NEXT: [[TMP63:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 7
// CHECK-NEXT: store ptr null, ptr [[TMP63]], align 8 // CHECK-NEXT: store ptr null, ptr [[TMP63]], align 8
// CHECK-NEXT: [[TMP64:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 7 // CHECK-NEXT: [[TMP64:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 8
// CHECK-NEXT: store ptr null, ptr [[TMP64]], align 8 // CHECK-NEXT: store i64 0, ptr [[TMP64]], align 8
// CHECK-NEXT: [[TMP65:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS16]], i32 0, i32 8 // CHECK-NEXT: [[TMP65:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPiET_S1__l129.region_id, ptr [[KERNEL_ARGS16]])
// CHECK-NEXT: store i64 0, ptr [[TMP65]], align 8 // CHECK-NEXT: [[TMP66:%.*]] = icmp ne i32 [[TMP65]], 0
// CHECK-NEXT: [[TMP66:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 -1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPiET_S1__l129.region_id, ptr [[KERNEL_ARGS16]]) // CHECK-NEXT: br i1 [[TMP66]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK-NEXT: [[TMP67:%.*]] = icmp ne i32 [[TMP66]], 0
// CHECK-NEXT: br i1 [[TMP67]], label [[OMP_OFFLOAD_FAILED17:%.*]], label [[OMP_OFFLOAD_CONT18:%.*]]
// CHECK: omp_offload.failed17: // CHECK: omp_offload.failed17:
// CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPiET_S1__l129(ptr [[H]]) #[[ATTR5]] // CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIPiET_S1__l129(ptr [[H]]) #[[ATTR5]]
// CHECK-NEXT: br label [[OMP_OFFLOAD_CONT18]] // CHECK-NEXT: br label [[OMP_OFFLOAD_CONT18]]
// CHECK: omp_offload.cont18: // CHECK: omp_offload.cont18:
// CHECK-NEXT: ret ptr null // CHECK-NEXT: ret ptr null
// //
// //
// CHECK-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_S0__l123 // CHECK-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_S0__l123
▲ Show 20 Lines • Show All 528 Lines • Show Last 20 Lines

clang/test/OpenMP/target_parallel_debug_codegen.cpp

Show First 20 Lines • Show All 89 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP]], align 8, !dbg [[DBG47]] // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP]], align 8, !dbg [[DBG47]]
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[B_ADDR]], align 8, !dbg [[DBG47]] // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[B_ADDR]], align 8, !dbg [[DBG47]]
// CHECK1-NEXT: store ptr [[TMP3]], ptr [[_TMP1]], align 8, !dbg [[DBG47]] // CHECK1-NEXT: store ptr [[TMP3]], ptr [[_TMP1]], align 8, !dbg [[DBG47]]
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[_TMP1]], align 8, !dbg [[DBG47]] // CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[_TMP1]], align 8, !dbg [[DBG47]]
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr addrspace(1), ptr [[BB_ADDR]], align 8, !dbg [[DBG47]] // CHECK1-NEXT: [[TMP5:%.*]] = load ptr addrspace(1), ptr [[BB_ADDR]], align 8, !dbg [[DBG47]]
// CHECK1-NEXT: [[TMP6:%.*]] = addrspacecast ptr addrspace(1) [[TMP5]] to ptr, !dbg [[DBG47]] // CHECK1-NEXT: [[TMP6:%.*]] = addrspacecast ptr addrspace(1) [[TMP5]] to ptr, !dbg [[DBG47]]
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[_TMP2]], align 8, !dbg [[DBG47]] // CHECK1-NEXT: store ptr [[TMP6]], ptr [[_TMP2]], align 8, !dbg [[DBG47]]
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[_TMP2]], align 8, !dbg [[DBG47]] // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[_TMP2]], align 8, !dbg [[DBG47]]
// CHECK1-NEXT: [[TMP8:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true), !dbg [[DBG47]] // CHECK1-NEXT: [[TMP8:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false), !dbg [[DBG47]]
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP8]], -1, !dbg [[DBG47]] // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP8]], -1, !dbg [[DBG47]]
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]], !dbg [[DBG47]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]], !dbg [[DBG47]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP9:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB3:[0-9]+]]) // CHECK1-NEXT: [[TMP9:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB3:[0-9]+]])
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG48:![0-9]+]] // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG48:![0-9]+]]
// CHECK1-NEXT: store i32 [[TMP10]], ptr [[A_CASTED]], align 4, !dbg [[DBG48]] // CHECK1-NEXT: store i32 [[TMP10]], ptr [[A_CASTED]], align 4, !dbg [[DBG48]]
// CHECK1-NEXT: [[TMP11:%.*]] = load i64, ptr [[A_CASTED]], align 8, !dbg [[DBG48]] // CHECK1-NEXT: [[TMP11:%.*]] = load i64, ptr [[A_CASTED]], align 8, !dbg [[DBG48]]
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0, !dbg [[DBG48]] // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0, !dbg [[DBG48]]
// CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP12]], align 8, !dbg [[DBG48]] // CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP12]], align 8, !dbg [[DBG48]]
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1, !dbg [[DBG48]] // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1, !dbg [[DBG48]]
// CHECK1-NEXT: [[TMP15:%.*]] = inttoptr i64 [[TMP11]] to ptr, !dbg [[DBG48]] // CHECK1-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP11]] to ptr, !dbg [[DBG48]]
// CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8, !dbg [[DBG48]] // CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8, !dbg [[DBG48]]
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2, !dbg [[DBG48]] // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2, !dbg [[DBG48]]
// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP16]], align 8, !dbg [[DBG48]] // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP15]], align 8, !dbg [[DBG48]]
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3, !dbg [[DBG48]] // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3, !dbg [[DBG48]]
// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP18]], align 8, !dbg [[DBG48]] // CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP16]], align 8, !dbg [[DBG48]]
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB3]], i32 [[TMP9]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4), !dbg [[DBG48]] // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB3]], i32 [[TMP9]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4), !dbg [[DBG48]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB5:[0-9]+]], i8 2, i1 true), !dbg [[DBG49:![0-9]+]] // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB5:[0-9]+]], i8 2), !dbg [[DBG49:![0-9]+]]
// CHECK1-NEXT: ret void, !dbg [[DBG51:![0-9]+]] // CHECK1-NEXT: ret void, !dbg [[DBG51:![0-9]+]]
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void, !dbg [[DBG47]] // CHECK1-NEXT: ret void, !dbg [[DBG47]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined___debug__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined___debug__
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr addrspace(1) noalias noundef [[C:%.*]], i32 noundef [[A:%.*]], ptr noalias noundef [[B:%.*]], ptr addrspace(1) noalias noundef [[BB:%.*]]) #[[ATTR0]] !dbg [[DBG52:![0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr addrspace(1) noalias noundef [[C:%.*]], i32 noundef [[A:%.*]], ptr noalias noundef [[B:%.*]], ptr addrspace(1) noalias noundef [[BB:%.*]]) #[[ATTR0]] !dbg [[DBG52:![0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 46 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[H]], metadata [[META79:![0-9]+]], metadata !DIExpression()), !dbg [[DBG80:![0-9]+]] // CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[H]], metadata [[META79:![0-9]+]], metadata !DIExpression()), !dbg [[DBG80:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[B3]], i64 0, i64 1, !dbg [[DBG81:![0-9]+]] // CHECK1-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[B3]], i64 0, i64 1, !dbg [[DBG81:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX6]], i64 0, i64 1, !dbg [[DBG81]] // CHECK1-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX6]], i64 0, i64 1, !dbg [[DBG81]]
// CHECK1-NEXT: store ptr [[ARRAYIDX7]], ptr [[H]], align 8, !dbg [[DBG80]] // CHECK1-NEXT: store ptr [[ARRAYIDX7]], ptr [[H]], align 8, !dbg [[DBG80]]
// CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[D]], metadata [[META82:![0-9]+]], metadata !DIExpression()), !dbg [[DBG83:![0-9]+]] // CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[D]], metadata [[META82:![0-9]+]], metadata !DIExpression()), !dbg [[DBG83:![0-9]+]]
// CHECK1-NEXT: store i32 15, ptr [[D]], align 4, !dbg [[DBG83]] // CHECK1-NEXT: store i32 15, ptr [[D]], align 4, !dbg [[DBG83]]
// CHECK1-NEXT: store i32 5, ptr [[A_ADDR]], align 4, !dbg [[DBG84:![0-9]+]] // CHECK1-NEXT: store i32 5, ptr [[A_ADDR]], align 4, !dbg [[DBG84:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[B3]], i64 0, i64 0, !dbg [[DBG85:![0-9]+]] // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[B3]], i64 0, i64 0, !dbg [[DBG85:![0-9]+]]
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG86:![0-9]+]] // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG86:![0-9]+]]
// CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP10]] to i64, !dbg [[DBG85]] // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP8]] to i64, !dbg [[DBG85]]
// CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX8]], i64 0, i64 [[IDXPROM]], !dbg [[DBG85]] // CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX8]], i64 0, i64 [[IDXPROM]], !dbg [[DBG85]]
// CHECK1-NEXT: store i32 10, ptr [[ARRAYIDX9]], align 4, !dbg [[DBG87:![0-9]+]] // CHECK1-NEXT: store i32 10, ptr [[ARRAYIDX9]], align 4, !dbg [[DBG87:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds [10 x [10 x [10 x i32]]], ptr [[TMP2]], i64 0, i64 0, !dbg [[DBG88:![0-9]+]] // CHECK1-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds [10 x [10 x [10 x i32]]], ptr [[TMP2]], i64 0, i64 0, !dbg [[DBG88:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[ARRAYIDX10]], i64 0, i64 0, !dbg [[DBG88]] // CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[ARRAYIDX10]], i64 0, i64 0, !dbg [[DBG88]]
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG89:![0-9]+]] // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG89:![0-9]+]]
// CHECK1-NEXT: [[IDXPROM12:%.*]] = sext i32 [[TMP11]] to i64, !dbg [[DBG88]] // CHECK1-NEXT: [[IDXPROM12:%.*]] = sext i32 [[TMP9]] to i64, !dbg [[DBG88]]
// CHECK1-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX11]], i64 0, i64 [[IDXPROM12]], !dbg [[DBG88]] // CHECK1-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX11]], i64 0, i64 [[IDXPROM12]], !dbg [[DBG88]]
// CHECK1-NEXT: store i32 11, ptr [[ARRAYIDX13]], align 4, !dbg [[DBG90:![0-9]+]] // CHECK1-NEXT: store i32 11, ptr [[ARRAYIDX13]], align 4, !dbg [[DBG90:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds [10 x [10 x [10 x i32]]], ptr [[TMP2]], i64 0, i64 0, !dbg [[DBG91:![0-9]+]] // CHECK1-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds [10 x [10 x [10 x i32]]], ptr [[TMP2]], i64 0, i64 0, !dbg [[DBG91:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX15:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[ARRAYIDX14]], i64 0, i64 0, !dbg [[DBG91]] // CHECK1-NEXT: [[ARRAYIDX15:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[ARRAYIDX14]], i64 0, i64 0, !dbg [[DBG91]]
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG92:![0-9]+]] // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG92:![0-9]+]]
// CHECK1-NEXT: [[IDXPROM16:%.*]] = sext i32 [[TMP12]] to i64, !dbg [[DBG91]] // CHECK1-NEXT: [[IDXPROM16:%.*]] = sext i32 [[TMP10]] to i64, !dbg [[DBG91]]
// CHECK1-NEXT: [[ARRAYIDX17:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX15]], i64 0, i64 [[IDXPROM16]], !dbg [[DBG91]] // CHECK1-NEXT: [[ARRAYIDX17:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX15]], i64 0, i64 [[IDXPROM16]], !dbg [[DBG91]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[ARRAYIDX17]], align 4, !dbg [[DBG91]] // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[ARRAYIDX17]], align 4, !dbg [[DBG91]]
// CHECK1-NEXT: [[ARRAYIDX18:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[B3]], i64 0, i64 0, !dbg [[DBG93:![0-9]+]] // CHECK1-NEXT: [[ARRAYIDX18:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[B3]], i64 0, i64 0, !dbg [[DBG93:![0-9]+]]
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG94:![0-9]+]] // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG94:![0-9]+]]
// CHECK1-NEXT: [[IDXPROM19:%.*]] = sext i32 [[TMP14]] to i64, !dbg [[DBG93]] // CHECK1-NEXT: [[IDXPROM19:%.*]] = sext i32 [[TMP12]] to i64, !dbg [[DBG93]]
// CHECK1-NEXT: [[ARRAYIDX20:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX18]], i64 0, i64 [[IDXPROM19]], !dbg [[DBG93]] // CHECK1-NEXT: [[ARRAYIDX20:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX18]], i64 0, i64 [[IDXPROM19]], !dbg [[DBG93]]
// CHECK1-NEXT: store i32 [[TMP13]], ptr [[ARRAYIDX20]], align 4, !dbg [[DBG95:![0-9]+]] // CHECK1-NEXT: store i32 [[TMP11]], ptr [[ARRAYIDX20]], align 4, !dbg [[DBG95:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX21:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[B3]], i64 0, i64 0, !dbg [[DBG96:![0-9]+]] // CHECK1-NEXT: [[ARRAYIDX21:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[B3]], i64 0, i64 0, !dbg [[DBG96:![0-9]+]]
// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG97:![0-9]+]] // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG97:![0-9]+]]
// CHECK1-NEXT: [[IDXPROM22:%.*]] = sext i32 [[TMP15]] to i64, !dbg [[DBG96]] // CHECK1-NEXT: [[IDXPROM22:%.*]] = sext i32 [[TMP13]] to i64, !dbg [[DBG96]]
// CHECK1-NEXT: [[ARRAYIDX23:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX21]], i64 0, i64 [[IDXPROM22]], !dbg [[DBG96]] // CHECK1-NEXT: [[ARRAYIDX23:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX21]], i64 0, i64 [[IDXPROM22]], !dbg [[DBG96]]
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[ARRAYIDX23]], align 4, !dbg [[DBG96]] // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[ARRAYIDX23]], align 4, !dbg [[DBG96]]
// CHECK1-NEXT: [[TMP17:%.*]] = load i8, ptr [[TMP7]], align 1, !dbg [[DBG98:![0-9]+]] // CHECK1-NEXT: [[TMP15:%.*]] = load i8, ptr [[TMP7]], align 1, !dbg [[DBG98:![0-9]+]]
// CHECK1-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP17]] to i1, !dbg [[DBG98]] // CHECK1-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP15]] to i1, !dbg [[DBG98]]
// CHECK1-NEXT: [[CONV:%.*]] = zext i1 [[TOBOOL]] to i32, !dbg [[DBG98]] // CHECK1-NEXT: [[CONV:%.*]] = zext i1 [[TOBOOL]] to i32, !dbg [[DBG98]]
// CHECK1-NEXT: [[OR:%.*]] = or i32 [[CONV]], [[TMP16]], !dbg [[DBG98]] // CHECK1-NEXT: [[OR:%.*]] = or i32 [[CONV]], [[TMP14]], !dbg [[DBG98]]
// CHECK1-NEXT: [[TOBOOL24:%.*]] = icmp ne i32 [[OR]], 0, !dbg [[DBG98]] // CHECK1-NEXT: [[TOBOOL24:%.*]] = icmp ne i32 [[OR]], 0, !dbg [[DBG98]]
// CHECK1-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL24]] to i8, !dbg [[DBG98]] // CHECK1-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL24]] to i8, !dbg [[DBG98]]
// CHECK1-NEXT: store i8 [[FROMBOOL]], ptr [[TMP7]], align 1, !dbg [[DBG98]] // CHECK1-NEXT: store i8 [[FROMBOOL]], ptr [[TMP7]], align 1, !dbg [[DBG98]]
// CHECK1-NEXT: ret void, !dbg [[DBG99:![0-9]+]] // CHECK1-NEXT: ret void, !dbg [[DBG99:![0-9]+]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[C:%.*]], i64 noundef [[A:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[B:%.*]], ptr noundef nonnull align 1 dereferenceable(1) [[BB:%.*]]) #[[ATTR0]] !dbg [[DBG100:![0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[C:%.*]], i64 noundef [[A:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[B:%.*]], ptr noundef nonnull align 1 dereferenceable(1) [[BB:%.*]]) #[[ATTR0]] !dbg [[DBG100:![0-9]+]] {
▲ Show 20 Lines • Show All 86 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr addrspace(1), ptr [[B_ADDR]], align 8, !dbg [[DBG137]] // CHECK1-NEXT: [[TMP3:%.*]] = load ptr addrspace(1), ptr [[B_ADDR]], align 8, !dbg [[DBG137]]
// CHECK1-NEXT: [[TMP4:%.*]] = addrspacecast ptr addrspace(1) [[TMP3]] to ptr, !dbg [[DBG137]] // CHECK1-NEXT: [[TMP4:%.*]] = addrspacecast ptr addrspace(1) [[TMP3]] to ptr, !dbg [[DBG137]]
// CHECK1-NEXT: store ptr [[TMP4]], ptr [[_TMP1]], align 8, !dbg [[DBG137]] // CHECK1-NEXT: store ptr [[TMP4]], ptr [[_TMP1]], align 8, !dbg [[DBG137]]
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP1]], align 8, !dbg [[DBG137]] // CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP1]], align 8, !dbg [[DBG137]]
// CHECK1-NEXT: [[TMP6:%.*]] = load ptr addrspace(1), ptr [[BB_ADDR]], align 8, !dbg [[DBG137]] // CHECK1-NEXT: [[TMP6:%.*]] = load ptr addrspace(1), ptr [[BB_ADDR]], align 8, !dbg [[DBG137]]
// CHECK1-NEXT: [[TMP7:%.*]] = addrspacecast ptr addrspace(1) [[TMP6]] to ptr, !dbg [[DBG137]] // CHECK1-NEXT: [[TMP7:%.*]] = addrspacecast ptr addrspace(1) [[TMP6]] to ptr, !dbg [[DBG137]]
// CHECK1-NEXT: store ptr [[TMP7]], ptr [[_TMP2]], align 8, !dbg [[DBG137]] // CHECK1-NEXT: store ptr [[TMP7]], ptr [[_TMP2]], align 8, !dbg [[DBG137]]
// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[_TMP2]], align 8, !dbg [[DBG137]] // CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[_TMP2]], align 8, !dbg [[DBG137]]
// CHECK1-NEXT: [[TMP9:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB7:[0-9]+]], i8 2, i1 false, i1 true), !dbg [[DBG137]] // CHECK1-NEXT: [[TMP9:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB7:[0-9]+]], i8 2, i1 false), !dbg [[DBG137]]
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP9]], -1, !dbg [[DBG137]] // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP9]], -1, !dbg [[DBG137]]
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]], !dbg [[DBG137]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]], !dbg [[DBG137]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP10:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB9:[0-9]+]]) // CHECK1-NEXT: [[TMP10:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB9:[0-9]+]])
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG138:![0-9]+]] // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG138:![0-9]+]]
// CHECK1-NEXT: store i32 [[TMP11]], ptr [[A_CASTED]], align 4, !dbg [[DBG138]] // CHECK1-NEXT: store i32 [[TMP11]], ptr [[A_CASTED]], align 4, !dbg [[DBG138]]
// CHECK1-NEXT: [[TMP12:%.*]] = load i64, ptr [[A_CASTED]], align 8, !dbg [[DBG138]] // CHECK1-NEXT: [[TMP12:%.*]] = load i64, ptr [[A_CASTED]], align 8, !dbg [[DBG138]]
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0, !dbg [[DBG138]] // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0, !dbg [[DBG138]]
// CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP13]], align 8, !dbg [[DBG138]] // CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP13]], align 8, !dbg [[DBG138]]
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1, !dbg [[DBG138]] // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1, !dbg [[DBG138]]
// CHECK1-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP12]] to ptr, !dbg [[DBG138]] // CHECK1-NEXT: [[TMP15:%.*]] = inttoptr i64 [[TMP12]] to ptr, !dbg [[DBG138]]
// CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8, !dbg [[DBG138]] // CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8, !dbg [[DBG138]]
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2, !dbg [[DBG138]] // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2, !dbg [[DBG138]]
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP17]], align 8, !dbg [[DBG138]] // CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP16]], align 8, !dbg [[DBG138]]
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3, !dbg [[DBG138]] // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3, !dbg [[DBG138]]
// CHECK1-NEXT: store ptr [[TMP8]], ptr [[TMP19]], align 8, !dbg [[DBG138]] // CHECK1-NEXT: store ptr [[TMP8]], ptr [[TMP17]], align 8, !dbg [[DBG138]]
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB9]], i32 [[TMP10]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__2, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4), !dbg [[DBG138]] // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB9]], i32 [[TMP10]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__2, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4), !dbg [[DBG138]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB11:[0-9]+]], i8 2, i1 true), !dbg [[DBG139:![0-9]+]] // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB11:[0-9]+]], i8 2), !dbg [[DBG139:![0-9]+]]
// CHECK1-NEXT: ret void, !dbg [[DBG141:![0-9]+]] // CHECK1-NEXT: ret void, !dbg [[DBG141:![0-9]+]]
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void, !dbg [[DBG137]] // CHECK1-NEXT: ret void, !dbg [[DBG137]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined___debug__1 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined___debug__1
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr addrspace(1) noalias noundef [[C:%.*]], i32 noundef [[A:%.*]], ptr addrspace(1) noalias noundef [[B:%.*]], ptr addrspace(1) noalias noundef [[BB:%.*]]) #[[ATTR0]] !dbg [[DBG142:![0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr addrspace(1) noalias noundef [[C:%.*]], i32 noundef [[A:%.*]], ptr addrspace(1) noalias noundef [[B:%.*]], ptr addrspace(1) noalias noundef [[BB:%.*]]) #[[ATTR0]] !dbg [[DBG142:![0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 174 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP6:%.*]] = load ptr addrspace(1), ptr [[B_ADDR]], align 8, !dbg [[DBG212]] // CHECK1-NEXT: [[TMP6:%.*]] = load ptr addrspace(1), ptr [[B_ADDR]], align 8, !dbg [[DBG212]]
// CHECK1-NEXT: [[TMP7:%.*]] = addrspacecast ptr addrspace(1) [[TMP6]] to ptr, !dbg [[DBG212]] // CHECK1-NEXT: [[TMP7:%.*]] = addrspacecast ptr addrspace(1) [[TMP6]] to ptr, !dbg [[DBG212]]
// CHECK1-NEXT: store ptr [[TMP7]], ptr [[_TMP2]], align 8, !dbg [[DBG212]] // CHECK1-NEXT: store ptr [[TMP7]], ptr [[_TMP2]], align 8, !dbg [[DBG212]]
// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[_TMP2]], align 8, !dbg [[DBG212]] // CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[_TMP2]], align 8, !dbg [[DBG212]]
// CHECK1-NEXT: [[TMP9:%.*]] = load ptr addrspace(1), ptr [[BB_ADDR]], align 8, !dbg [[DBG212]] // CHECK1-NEXT: [[TMP9:%.*]] = load ptr addrspace(1), ptr [[BB_ADDR]], align 8, !dbg [[DBG212]]
// CHECK1-NEXT: [[TMP10:%.*]] = addrspacecast ptr addrspace(1) [[TMP9]] to ptr, !dbg [[DBG212]] // CHECK1-NEXT: [[TMP10:%.*]] = addrspacecast ptr addrspace(1) [[TMP9]] to ptr, !dbg [[DBG212]]
// CHECK1-NEXT: store ptr [[TMP10]], ptr [[_TMP3]], align 8, !dbg [[DBG212]] // CHECK1-NEXT: store ptr [[TMP10]], ptr [[_TMP3]], align 8, !dbg [[DBG212]]
// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[_TMP3]], align 8, !dbg [[DBG212]] // CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[_TMP3]], align 8, !dbg [[DBG212]]
// CHECK1-NEXT: [[TMP12:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB13:[0-9]+]], i8 2, i1 false, i1 true), !dbg [[DBG212]] // CHECK1-NEXT: [[TMP12:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB13:[0-9]+]], i8 2, i1 false), !dbg [[DBG212]]
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP12]], -1, !dbg [[DBG212]] // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP12]], -1, !dbg [[DBG212]]
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]], !dbg [[DBG212]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]], !dbg [[DBG212]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP13:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB15:[0-9]+]]) // CHECK1-NEXT: [[TMP13:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB15:[0-9]+]])
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0, !dbg [[DBG213:![0-9]+]] // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0, !dbg [[DBG213:![0-9]+]]
// CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP14]], align 8, !dbg [[DBG213]] // CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP14]], align 8, !dbg [[DBG213]]
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1, !dbg [[DBG213]] // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1, !dbg [[DBG213]]
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP16]], align 8, !dbg [[DBG213]] // CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP15]], align 8, !dbg [[DBG213]]
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2, !dbg [[DBG213]] // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2, !dbg [[DBG213]]
// CHECK1-NEXT: store ptr [[TMP8]], ptr [[TMP18]], align 8, !dbg [[DBG213]] // CHECK1-NEXT: store ptr [[TMP8]], ptr [[TMP16]], align 8, !dbg [[DBG213]]
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3, !dbg [[DBG213]] // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3, !dbg [[DBG213]]
// CHECK1-NEXT: store ptr [[TMP11]], ptr [[TMP20]], align 8, !dbg [[DBG213]] // CHECK1-NEXT: store ptr [[TMP11]], ptr [[TMP17]], align 8, !dbg [[DBG213]]
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB15]], i32 [[TMP13]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__4, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4), !dbg [[DBG213]] // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB15]], i32 [[TMP13]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__4, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4), !dbg [[DBG213]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB17:[0-9]+]], i8 2, i1 true), !dbg [[DBG214:![0-9]+]] // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB17:[0-9]+]], i8 2), !dbg [[DBG214:![0-9]+]]
// CHECK1-NEXT: ret void, !dbg [[DBG216:![0-9]+]] // CHECK1-NEXT: ret void, !dbg [[DBG216:![0-9]+]]
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void, !dbg [[DBG212]] // CHECK1-NEXT: ret void, !dbg [[DBG212]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined___debug__3 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined___debug__3
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr addrspace(1) noalias noundef [[C:%.*]], ptr addrspace(1) noalias noundef [[A:%.*]], ptr addrspace(1) noalias noundef [[B:%.*]], ptr addrspace(1) noalias noundef [[BB:%.*]]) #[[ATTR0]] !dbg [[DBG217:![0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr addrspace(1) noalias noundef [[C:%.*]], ptr addrspace(1) noalias noundef [[A:%.*]], ptr addrspace(1) noalias noundef [[B:%.*]], ptr addrspace(1) noalias noundef [[BB:%.*]]) #[[ATTR0]] !dbg [[DBG217:![0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 158 Lines • Show Last 20 Lines

clang/test/OpenMP/target_parallel_for_debug_codegen.cpp

Show First 20 Lines • Show All 83 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP]], align 8, !dbg [[DBG41]] // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[TMP]], align 8, !dbg [[DBG41]]
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[B_ADDR]], align 8, !dbg [[DBG41]] // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[B_ADDR]], align 8, !dbg [[DBG41]]
// CHECK1-NEXT: store ptr [[TMP3]], ptr [[_TMP1]], align 8, !dbg [[DBG41]] // CHECK1-NEXT: store ptr [[TMP3]], ptr [[_TMP1]], align 8, !dbg [[DBG41]]
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[_TMP1]], align 8, !dbg [[DBG41]] // CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[_TMP1]], align 8, !dbg [[DBG41]]
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr addrspace(1), ptr [[BB_ADDR]], align 8, !dbg [[DBG41]] // CHECK1-NEXT: [[TMP5:%.*]] = load ptr addrspace(1), ptr [[BB_ADDR]], align 8, !dbg [[DBG41]]
// CHECK1-NEXT: [[TMP6:%.*]] = addrspacecast ptr addrspace(1) [[TMP5]] to ptr, !dbg [[DBG41]] // CHECK1-NEXT: [[TMP6:%.*]] = addrspacecast ptr addrspace(1) [[TMP5]] to ptr, !dbg [[DBG41]]
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[_TMP2]], align 8, !dbg [[DBG41]] // CHECK1-NEXT: store ptr [[TMP6]], ptr [[_TMP2]], align 8, !dbg [[DBG41]]
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[_TMP2]], align 8, !dbg [[DBG41]] // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[_TMP2]], align 8, !dbg [[DBG41]]
// CHECK1-NEXT: [[TMP8:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true), !dbg [[DBG41]] // CHECK1-NEXT: [[TMP8:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB1:[0-9]+]], i8 2, i1 false), !dbg [[DBG41]]
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP8]], -1, !dbg [[DBG41]] // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP8]], -1, !dbg [[DBG41]]
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]], !dbg [[DBG41]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]], !dbg [[DBG41]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP9:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB6:[0-9]+]]) // CHECK1-NEXT: [[TMP9:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB6:[0-9]+]])
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG42:![0-9]+]] // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG42:![0-9]+]]
// CHECK1-NEXT: store i32 [[TMP10]], ptr [[A_CASTED]], align 4, !dbg [[DBG42]] // CHECK1-NEXT: store i32 [[TMP10]], ptr [[A_CASTED]], align 4, !dbg [[DBG42]]
// CHECK1-NEXT: [[TMP11:%.*]] = load i64, ptr [[A_CASTED]], align 8, !dbg [[DBG42]] // CHECK1-NEXT: [[TMP11:%.*]] = load i64, ptr [[A_CASTED]], align 8, !dbg [[DBG42]]
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0, !dbg [[DBG42]] // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0, !dbg [[DBG42]]
// CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP12]], align 8, !dbg [[DBG42]] // CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP12]], align 8, !dbg [[DBG42]]
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1, !dbg [[DBG42]] // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1, !dbg [[DBG42]]
// CHECK1-NEXT: [[TMP15:%.*]] = inttoptr i64 [[TMP11]] to ptr, !dbg [[DBG42]] // CHECK1-NEXT: [[TMP14:%.*]] = inttoptr i64 [[TMP11]] to ptr, !dbg [[DBG42]]
// CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8, !dbg [[DBG42]] // CHECK1-NEXT: store ptr [[TMP14]], ptr [[TMP13]], align 8, !dbg [[DBG42]]
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2, !dbg [[DBG42]] // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2, !dbg [[DBG42]]
// CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP16]], align 8, !dbg [[DBG42]] // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP15]], align 8, !dbg [[DBG42]]
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3, !dbg [[DBG42]] // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3, !dbg [[DBG42]]
// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP18]], align 8, !dbg [[DBG42]] // CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP16]], align 8, !dbg [[DBG42]]
// CHECK1-NEXT: [[TMP19:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !dbg [[DBG43:![0-9]+]] // CHECK1-NEXT: [[TMP17:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1, !dbg [[DBG43:![0-9]+]]
// CHECK1-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP19]] to i1, !dbg [[DBG43]] // CHECK1-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP17]] to i1, !dbg [[DBG43]]
// CHECK1-NEXT: [[TMP20:%.*]] = zext i1 [[TOBOOL]] to i32, !dbg [[DBG42]] // CHECK1-NEXT: [[TMP18:%.*]] = zext i1 [[TOBOOL]] to i32, !dbg [[DBG42]]
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB6]], i32 [[TMP9]], i32 [[TMP20]], i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4), !dbg [[DBG42]] // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB6]], i32 [[TMP9]], i32 [[TMP18]], i32 -1, i32 -1, ptr @__omp_outlined__, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4), !dbg [[DBG42]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB8:[0-9]+]], i8 2, i1 true), !dbg [[DBG45:![0-9]+]] // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB8:[0-9]+]], i8 2), !dbg [[DBG45:![0-9]+]]
// CHECK1-NEXT: ret void, !dbg [[DBG46:![0-9]+]] // CHECK1-NEXT: ret void, !dbg [[DBG46:![0-9]+]]
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void, !dbg [[DBG41]] // CHECK1-NEXT: ret void, !dbg [[DBG41]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined___debug__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined___debug__
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr addrspace(1) noalias noundef [[C:%.*]], i32 noundef [[A:%.*]], ptr noalias noundef [[B:%.*]], ptr addrspace(1) noalias noundef [[BB:%.*]]) #[[ATTR0]] !dbg [[DBG47:![0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr addrspace(1) noalias noundef [[C:%.*]], i32 noundef [[A:%.*]], ptr noalias noundef [[B:%.*]], ptr addrspace(1) noalias noundef [[BB:%.*]]) #[[ATTR0]] !dbg [[DBG47:![0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[DOTOMP_STRIDE]], metadata [[META70:![0-9]+]], metadata !DIExpression()), !dbg [[DBG55]] // CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[DOTOMP_STRIDE]], metadata [[META70:![0-9]+]], metadata !DIExpression()), !dbg [[DBG55]]
// CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[DOTOMP_IS_LAST]], metadata [[META71:![0-9]+]], metadata !DIExpression()), !dbg [[DBG55]] // CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[DOTOMP_IS_LAST]], metadata [[META71:![0-9]+]], metadata !DIExpression()), !dbg [[DBG55]]
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[B4]], metadata [[META72:![0-9]+]], metadata !DIExpression()), !dbg [[DBG55]] // CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[B4]], metadata [[META72:![0-9]+]], metadata !DIExpression()), !dbg [[DBG55]]
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[B4]], ptr align 4 [[TMP4]], i64 400, i1 false), !dbg [[DBG65]] // CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[B4]], ptr align 4 [[TMP4]], i64 400, i1 false), !dbg [[DBG65]]
// CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[I]], metadata [[META73:![0-9]+]], metadata !DIExpression()), !dbg [[DBG55]] // CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[I]], metadata [[META73:![0-9]+]], metadata !DIExpression()), !dbg [[DBG55]]
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8, !dbg [[DBG65]] // CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8, !dbg [[DBG65]]
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4, !dbg [[DBG65]] // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4, !dbg [[DBG65]]
// CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP11]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1), !dbg [[DBG74:![0-9]+]] // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB3:[0-9]+]], i32 [[TMP9]], i32 33, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1), !dbg [[DBG74:![0-9]+]]
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND:%.*]], !dbg [[DBG65]] // CHECK1-NEXT: br label [[OMP_DISPATCH_COND:%.*]], !dbg [[DBG65]]
// CHECK1: omp.dispatch.cond: // CHECK1: omp.dispatch.cond:
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP12]], 9, !dbg [[DBG68]] // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP10]], 9, !dbg [[DBG68]]
// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]], !dbg [[DBG68]] // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]], !dbg [[DBG68]]
// CHECK1: cond.true: // CHECK1: cond.true:
// CHECK1-NEXT: br label [[COND_END:%.*]], !dbg [[DBG68]] // CHECK1-NEXT: br label [[COND_END:%.*]], !dbg [[DBG68]]
// CHECK1: cond.false: // CHECK1: cond.false:
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: br label [[COND_END]], !dbg [[DBG68]] // CHECK1-NEXT: br label [[COND_END]], !dbg [[DBG68]]
// CHECK1: cond.end: // CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ], !dbg [[DBG68]] // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ], !dbg [[DBG68]]
// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: store i32 [[TMP12]], ptr [[DOTOMP_IV]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]], !dbg [[DBG65]] // CHECK1-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]], !dbg [[DBG65]]
// CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]], !dbg [[DBG65]] // CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]], !dbg [[DBG65]]
// CHECK1: omp.dispatch.body: // CHECK1: omp.dispatch.body:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]], !dbg [[DBG65]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]], !dbg [[DBG65]]
// CHECK1: omp.inner.for.cond: // CHECK1: omp.inner.for.cond:
// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]], !dbg [[DBG65]] // CHECK1-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]], !dbg [[DBG65]]
// CHECK1-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]], !dbg [[DBG65]] // CHECK1-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]], !dbg [[DBG65]]
// CHECK1: omp.inner.for.body: // CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1, !dbg [[DBG75:![0-9]+]] // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1, !dbg [[DBG75:![0-9]+]]
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]], !dbg [[DBG75]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]], !dbg [[DBG75]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !dbg [[DBG75]] // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !dbg [[DBG75]]
// CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[F]], metadata [[META76:![0-9]+]], metadata !DIExpression()), !dbg [[DBG79:![0-9]+]] // CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[F]], metadata [[META76:![0-9]+]], metadata !DIExpression()), !dbg [[DBG79:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [10 x [10 x i32]]], ptr [[TMP2]], i64 0, i64 1, !dbg [[DBG80:![0-9]+]] // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [10 x [10 x i32]]], ptr [[TMP2]], i64 0, i64 1, !dbg [[DBG80:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[ARRAYIDX]], i64 0, i64 1, !dbg [[DBG80]] // CHECK1-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[ARRAYIDX]], i64 0, i64 1, !dbg [[DBG80]]
// CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX7]], i64 0, i64 1, !dbg [[DBG80]] // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX7]], i64 0, i64 1, !dbg [[DBG80]]
// CHECK1-NEXT: store ptr [[ARRAYIDX8]], ptr [[F]], align 8, !dbg [[DBG79]] // CHECK1-NEXT: store ptr [[ARRAYIDX8]], ptr [[F]], align 8, !dbg [[DBG79]]
// CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[G]], metadata [[META81:![0-9]+]], metadata !DIExpression()), !dbg [[DBG82:![0-9]+]] // CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[G]], metadata [[META81:![0-9]+]], metadata !DIExpression()), !dbg [[DBG82:![0-9]+]]
// CHECK1-NEXT: store ptr [[A_ADDR]], ptr [[G]], align 8, !dbg [[DBG82]] // CHECK1-NEXT: store ptr [[A_ADDR]], ptr [[G]], align 8, !dbg [[DBG82]]
// CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[H]], metadata [[META83:![0-9]+]], metadata !DIExpression()), !dbg [[DBG84:![0-9]+]] // CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[H]], metadata [[META83:![0-9]+]], metadata !DIExpression()), !dbg [[DBG84:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[B4]], i64 0, i64 1, !dbg [[DBG85:![0-9]+]] // CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[B4]], i64 0, i64 1, !dbg [[DBG85:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX9]], i64 0, i64 1, !dbg [[DBG85]] // CHECK1-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX9]], i64 0, i64 1, !dbg [[DBG85]]
// CHECK1-NEXT: store ptr [[ARRAYIDX10]], ptr [[H]], align 8, !dbg [[DBG84]] // CHECK1-NEXT: store ptr [[ARRAYIDX10]], ptr [[H]], align 8, !dbg [[DBG84]]
// CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[D]], metadata [[META86:![0-9]+]], metadata !DIExpression()), !dbg [[DBG87:![0-9]+]] // CHECK1-NEXT: call void @llvm.dbg.declare(metadata ptr [[D]], metadata [[META86:![0-9]+]], metadata !DIExpression()), !dbg [[DBG87:![0-9]+]]
// CHECK1-NEXT: store i32 15, ptr [[D]], align 4, !dbg [[DBG87]] // CHECK1-NEXT: store i32 15, ptr [[D]], align 4, !dbg [[DBG87]]
// CHECK1-NEXT: store i32 5, ptr [[A_ADDR]], align 4, !dbg [[DBG88:![0-9]+]] // CHECK1-NEXT: store i32 5, ptr [[A_ADDR]], align 4, !dbg [[DBG88:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[B4]], i64 0, i64 0, !dbg [[DBG89:![0-9]+]] // CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[B4]], i64 0, i64 0, !dbg [[DBG89:![0-9]+]]
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG90:![0-9]+]] // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG90:![0-9]+]]
// CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP20]] to i64, !dbg [[DBG89]] // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP18]] to i64, !dbg [[DBG89]]
// CHECK1-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX11]], i64 0, i64 [[IDXPROM]], !dbg [[DBG89]] // CHECK1-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX11]], i64 0, i64 [[IDXPROM]], !dbg [[DBG89]]
// CHECK1-NEXT: store i32 10, ptr [[ARRAYIDX12]], align 4, !dbg [[DBG91:![0-9]+]] // CHECK1-NEXT: store i32 10, ptr [[ARRAYIDX12]], align 4, !dbg [[DBG91:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds [10 x [10 x [10 x i32]]], ptr [[TMP2]], i64 0, i64 0, !dbg [[DBG92:![0-9]+]] // CHECK1-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds [10 x [10 x [10 x i32]]], ptr [[TMP2]], i64 0, i64 0, !dbg [[DBG92:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[ARRAYIDX13]], i64 0, i64 0, !dbg [[DBG92]] // CHECK1-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[ARRAYIDX13]], i64 0, i64 0, !dbg [[DBG92]]
// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG93:![0-9]+]] // CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG93:![0-9]+]]
// CHECK1-NEXT: [[IDXPROM15:%.*]] = sext i32 [[TMP21]] to i64, !dbg [[DBG92]] // CHECK1-NEXT: [[IDXPROM15:%.*]] = sext i32 [[TMP19]] to i64, !dbg [[DBG92]]
// CHECK1-NEXT: [[ARRAYIDX16:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX14]], i64 0, i64 [[IDXPROM15]], !dbg [[DBG92]] // CHECK1-NEXT: [[ARRAYIDX16:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX14]], i64 0, i64 [[IDXPROM15]], !dbg [[DBG92]]
// CHECK1-NEXT: store i32 11, ptr [[ARRAYIDX16]], align 4, !dbg [[DBG94:![0-9]+]] // CHECK1-NEXT: store i32 11, ptr [[ARRAYIDX16]], align 4, !dbg [[DBG94:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX17:%.*]] = getelementptr inbounds [10 x [10 x [10 x i32]]], ptr [[TMP2]], i64 0, i64 0, !dbg [[DBG95:![0-9]+]] // CHECK1-NEXT: [[ARRAYIDX17:%.*]] = getelementptr inbounds [10 x [10 x [10 x i32]]], ptr [[TMP2]], i64 0, i64 0, !dbg [[DBG95:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX18:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[ARRAYIDX17]], i64 0, i64 0, !dbg [[DBG95]] // CHECK1-NEXT: [[ARRAYIDX18:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[ARRAYIDX17]], i64 0, i64 0, !dbg [[DBG95]]
// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG96:![0-9]+]] // CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG96:![0-9]+]]
// CHECK1-NEXT: [[IDXPROM19:%.*]] = sext i32 [[TMP22]] to i64, !dbg [[DBG95]] // CHECK1-NEXT: [[IDXPROM19:%.*]] = sext i32 [[TMP20]] to i64, !dbg [[DBG95]]
// CHECK1-NEXT: [[ARRAYIDX20:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX18]], i64 0, i64 [[IDXPROM19]], !dbg [[DBG95]] // CHECK1-NEXT: [[ARRAYIDX20:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX18]], i64 0, i64 [[IDXPROM19]], !dbg [[DBG95]]
// CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[ARRAYIDX20]], align 4, !dbg [[DBG95]] // CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[ARRAYIDX20]], align 4, !dbg [[DBG95]]
// CHECK1-NEXT: [[ARRAYIDX21:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[B4]], i64 0, i64 0, !dbg [[DBG97:![0-9]+]] // CHECK1-NEXT: [[ARRAYIDX21:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[B4]], i64 0, i64 0, !dbg [[DBG97:![0-9]+]]
// CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG98:![0-9]+]] // CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG98:![0-9]+]]
// CHECK1-NEXT: [[IDXPROM22:%.*]] = sext i32 [[TMP24]] to i64, !dbg [[DBG97]] // CHECK1-NEXT: [[IDXPROM22:%.*]] = sext i32 [[TMP22]] to i64, !dbg [[DBG97]]
// CHECK1-NEXT: [[ARRAYIDX23:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX21]], i64 0, i64 [[IDXPROM22]], !dbg [[DBG97]] // CHECK1-NEXT: [[ARRAYIDX23:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX21]], i64 0, i64 [[IDXPROM22]], !dbg [[DBG97]]
// CHECK1-NEXT: store i32 [[TMP23]], ptr [[ARRAYIDX23]], align 4, !dbg [[DBG99:![0-9]+]] // CHECK1-NEXT: store i32 [[TMP21]], ptr [[ARRAYIDX23]], align 4, !dbg [[DBG99:![0-9]+]]
// CHECK1-NEXT: [[ARRAYIDX24:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[B4]], i64 0, i64 0, !dbg [[DBG100:![0-9]+]] // CHECK1-NEXT: [[ARRAYIDX24:%.*]] = getelementptr inbounds [10 x [10 x i32]], ptr [[B4]], i64 0, i64 0, !dbg [[DBG100:![0-9]+]]
// CHECK1-NEXT: [[TMP25:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG101:![0-9]+]] // CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG101:![0-9]+]]
// CHECK1-NEXT: [[IDXPROM25:%.*]] = sext i32 [[TMP25]] to i64, !dbg [[DBG100]] // CHECK1-NEXT: [[IDXPROM25:%.*]] = sext i32 [[TMP23]] to i64, !dbg [[DBG100]]
// CHECK1-NEXT: [[ARRAYIDX26:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX24]], i64 0, i64 [[IDXPROM25]], !dbg [[DBG100]] // CHECK1-NEXT: [[ARRAYIDX26:%.*]] = getelementptr inbounds [10 x i32], ptr [[ARRAYIDX24]], i64 0, i64 [[IDXPROM25]], !dbg [[DBG100]]
// CHECK1-NEXT: [[TMP26:%.*]] = load i32, ptr [[ARRAYIDX26]], align 4, !dbg [[DBG100]] // CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[ARRAYIDX26]], align 4, !dbg [[DBG100]]
// CHECK1-NEXT: [[TMP27:%.*]] = load i8, ptr [[TMP7]], align 1, !dbg [[DBG102:![0-9]+]] // CHECK1-NEXT: [[TMP25:%.*]] = load i8, ptr [[TMP7]], align 1, !dbg [[DBG102:![0-9]+]]
// CHECK1-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP27]] to i1, !dbg [[DBG102]] // CHECK1-NEXT: [[TOBOOL:%.*]] = trunc i8 [[TMP25]] to i1, !dbg [[DBG102]]
// CHECK1-NEXT: [[CONV:%.*]] = zext i1 [[TOBOOL]] to i32, !dbg [[DBG102]] // CHECK1-NEXT: [[CONV:%.*]] = zext i1 [[TOBOOL]] to i32, !dbg [[DBG102]]
// CHECK1-NEXT: [[OR:%.*]] = or i32 [[CONV]], [[TMP26]], !dbg [[DBG102]] // CHECK1-NEXT: [[OR:%.*]] = or i32 [[CONV]], [[TMP24]], !dbg [[DBG102]]
// CHECK1-NEXT: [[TOBOOL27:%.*]] = icmp ne i32 [[OR]], 0, !dbg [[DBG102]] // CHECK1-NEXT: [[TOBOOL27:%.*]] = icmp ne i32 [[OR]], 0, !dbg [[DBG102]]
// CHECK1-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL27]] to i8, !dbg [[DBG102]] // CHECK1-NEXT: [[FROMBOOL:%.*]] = zext i1 [[TOBOOL27]] to i8, !dbg [[DBG102]]
// CHECK1-NEXT: store i8 [[FROMBOOL]], ptr [[TMP7]], align 1, !dbg [[DBG102]] // CHECK1-NEXT: store i8 [[FROMBOOL]], ptr [[TMP7]], align 1, !dbg [[DBG102]]
// CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]], !dbg [[DBG103:![0-9]+]] // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]], !dbg [[DBG103:![0-9]+]]
// CHECK1: omp.body.continue: // CHECK1: omp.body.continue:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]], !dbg [[DBG74]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]], !dbg [[DBG74]]
// CHECK1: omp.inner.for.inc: // CHECK1: omp.inner.for.inc:
// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP28]], 1, !dbg [[DBG65]] // CHECK1-NEXT: [[ADD28:%.*]] = add nsw i32 [[TMP26]], 1, !dbg [[DBG65]]
// CHECK1-NEXT: store i32 [[ADD28]], ptr [[DOTOMP_IV]], align 4, !dbg [[DBG65]] // CHECK1-NEXT: store i32 [[ADD28]], ptr [[DOTOMP_IV]], align 4, !dbg [[DBG65]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !dbg [[DBG74]], !llvm.loop [[LOOP104:![0-9]+]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !dbg [[DBG74]], !llvm.loop [[LOOP104:![0-9]+]]
// CHECK1: omp.inner.for.end: // CHECK1: omp.inner.for.end:
// CHECK1-NEXT: br label [[OMP_DISPATCH_INC:%.*]], !dbg [[DBG74]] // CHECK1-NEXT: br label [[OMP_DISPATCH_INC:%.*]], !dbg [[DBG74]]
// CHECK1: omp.dispatch.inc: // CHECK1: omp.dispatch.inc:
// CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP29]], [[TMP30]], !dbg [[DBG65]] // CHECK1-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP27]], [[TMP28]], !dbg [[DBG65]]
// CHECK1-NEXT: store i32 [[ADD29]], ptr [[DOTOMP_LB]], align 4, !dbg [[DBG65]] // CHECK1-NEXT: store i32 [[ADD29]], ptr [[DOTOMP_LB]], align 4, !dbg [[DBG65]]
// CHECK1-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !dbg [[DBG68]] // CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !dbg [[DBG68]]
// CHECK1-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP31]], [[TMP32]], !dbg [[DBG65]] // CHECK1-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP29]], [[TMP30]], !dbg [[DBG65]]
// CHECK1-NEXT: store i32 [[ADD30]], ptr [[DOTOMP_UB]], align 4, !dbg [[DBG65]] // CHECK1-NEXT: store i32 [[ADD30]], ptr [[DOTOMP_UB]], align 4, !dbg [[DBG65]]
// CHECK1-NEXT: br label [[OMP_DISPATCH_COND]], !dbg [[DBG74]], !llvm.loop [[LOOP106:![0-9]+]] // CHECK1-NEXT: br label [[OMP_DISPATCH_COND]], !dbg [[DBG74]], !llvm.loop [[LOOP106:![0-9]+]]
// CHECK1: omp.dispatch.end: // CHECK1: omp.dispatch.end:
// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB5:[0-9]+]], i32 [[TMP11]]), !dbg [[DBG105:![0-9]+]] // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB5:[0-9]+]], i32 [[TMP9]]), !dbg [[DBG105:![0-9]+]]
// CHECK1-NEXT: ret void, !dbg [[DBG107:![0-9]+]] // CHECK1-NEXT: ret void, !dbg [[DBG107:![0-9]+]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__ // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined__
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[C:%.*]], i64 noundef [[A:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[B:%.*]], ptr noundef nonnull align 1 dereferenceable(1) [[BB:%.*]]) #[[ATTR0]] !dbg [[DBG108:![0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4000) [[C:%.*]], i64 noundef [[A:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[B:%.*]], ptr noundef nonnull align 1 dereferenceable(1) [[BB:%.*]]) #[[ATTR0]] !dbg [[DBG108:![0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
▲ Show 20 Lines • Show All 88 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr addrspace(1), ptr [[B_ADDR]], align 8, !dbg [[DBG146]] // CHECK1-NEXT: [[TMP3:%.*]] = load ptr addrspace(1), ptr [[B_ADDR]], align 8, !dbg [[DBG146]]
// CHECK1-NEXT: [[TMP4:%.*]] = addrspacecast ptr addrspace(1) [[TMP3]] to ptr, !dbg [[DBG146]] // CHECK1-NEXT: [[TMP4:%.*]] = addrspacecast ptr addrspace(1) [[TMP3]] to ptr, !dbg [[DBG146]]
// CHECK1-NEXT: store ptr [[TMP4]], ptr [[_TMP1]], align 8, !dbg [[DBG146]] // CHECK1-NEXT: store ptr [[TMP4]], ptr [[_TMP1]], align 8, !dbg [[DBG146]]
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP1]], align 8, !dbg [[DBG146]] // CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[_TMP1]], align 8, !dbg [[DBG146]]
// CHECK1-NEXT: [[TMP6:%.*]] = load ptr addrspace(1), ptr [[BB_ADDR]], align 8, !dbg [[DBG146]] // CHECK1-NEXT: [[TMP6:%.*]] = load ptr addrspace(1), ptr [[BB_ADDR]], align 8, !dbg [[DBG146]]
// CHECK1-NEXT: [[TMP7:%.*]] = addrspacecast ptr addrspace(1) [[TMP6]] to ptr, !dbg [[DBG146]] // CHECK1-NEXT: [[TMP7:%.*]] = addrspacecast ptr addrspace(1) [[TMP6]] to ptr, !dbg [[DBG146]]
// CHECK1-NEXT: store ptr [[TMP7]], ptr [[_TMP2]], align 8, !dbg [[DBG146]] // CHECK1-NEXT: store ptr [[TMP7]], ptr [[_TMP2]], align 8, !dbg [[DBG146]]
// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[_TMP2]], align 8, !dbg [[DBG146]] // CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[_TMP2]], align 8, !dbg [[DBG146]]
// CHECK1-NEXT: [[TMP9:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB10:[0-9]+]], i8 2, i1 false, i1 true), !dbg [[DBG146]] // CHECK1-NEXT: [[TMP9:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB10:[0-9]+]], i8 2, i1 false), !dbg [[DBG146]]
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP9]], -1, !dbg [[DBG146]] // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP9]], -1, !dbg [[DBG146]]
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]], !dbg [[DBG146]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]], !dbg [[DBG146]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP10:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB15:[0-9]+]]) // CHECK1-NEXT: [[TMP10:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB15:[0-9]+]])
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG147:![0-9]+]] // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[A_ADDR]], align 4, !dbg [[DBG147:![0-9]+]]
// CHECK1-NEXT: store i32 [[TMP11]], ptr [[A_CASTED]], align 4, !dbg [[DBG147]] // CHECK1-NEXT: store i32 [[TMP11]], ptr [[A_CASTED]], align 4, !dbg [[DBG147]]
// CHECK1-NEXT: [[TMP12:%.*]] = load i64, ptr [[A_CASTED]], align 8, !dbg [[DBG147]] // CHECK1-NEXT: [[TMP12:%.*]] = load i64, ptr [[A_CASTED]], align 8, !dbg [[DBG147]]
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0, !dbg [[DBG147]] // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0, !dbg [[DBG147]]
// CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP13]], align 8, !dbg [[DBG147]] // CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP13]], align 8, !dbg [[DBG147]]
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1, !dbg [[DBG147]] // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1, !dbg [[DBG147]]
// CHECK1-NEXT: [[TMP16:%.*]] = inttoptr i64 [[TMP12]] to ptr, !dbg [[DBG147]] // CHECK1-NEXT: [[TMP15:%.*]] = inttoptr i64 [[TMP12]] to ptr, !dbg [[DBG147]]
// CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP15]], align 8, !dbg [[DBG147]] // CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP14]], align 8, !dbg [[DBG147]]
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2, !dbg [[DBG147]] // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2, !dbg [[DBG147]]
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP17]], align 8, !dbg [[DBG147]] // CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP16]], align 8, !dbg [[DBG147]]
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3, !dbg [[DBG147]] // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3, !dbg [[DBG147]]
// CHECK1-NEXT: store ptr [[TMP8]], ptr [[TMP19]], align 8, !dbg [[DBG147]] // CHECK1-NEXT: store ptr [[TMP8]], ptr [[TMP17]], align 8, !dbg [[DBG147]]
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB15]], i32 [[TMP10]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__2, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4), !dbg [[DBG147]] // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB15]], i32 [[TMP10]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__2, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4), !dbg [[DBG147]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB17:[0-9]+]], i8 2, i1 true), !dbg [[DBG148:![0-9]+]] // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB17:[0-9]+]], i8 2), !dbg [[DBG148:![0-9]+]]
// CHECK1-NEXT: ret void, !dbg [[DBG150:![0-9]+]] // CHECK1-NEXT: ret void, !dbg [[DBG150:![0-9]+]]
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void, !dbg [[DBG146]] // CHECK1-NEXT: ret void, !dbg [[DBG146]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined___debug__1 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined___debug__1
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr addrspace(1) noalias noundef [[C:%.*]], i32 noundef [[A:%.*]], ptr addrspace(1) noalias noundef [[B:%.*]], ptr addrspace(1) noalias noundef [[BB:%.*]]) #[[ATTR0]] !dbg [[DBG151:![0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr addrspace(1) noalias noundef [[C:%.*]], i32 noundef [[A:%.*]], ptr addrspace(1) noalias noundef [[B:%.*]], ptr addrspace(1) noalias noundef [[BB:%.*]]) #[[ATTR0]] !dbg [[DBG151:![0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 247 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP6:%.*]] = load ptr addrspace(1), ptr [[B_ADDR]], align 8, !dbg [[DBG236]] // CHECK1-NEXT: [[TMP6:%.*]] = load ptr addrspace(1), ptr [[B_ADDR]], align 8, !dbg [[DBG236]]
// CHECK1-NEXT: [[TMP7:%.*]] = addrspacecast ptr addrspace(1) [[TMP6]] to ptr, !dbg [[DBG236]] // CHECK1-NEXT: [[TMP7:%.*]] = addrspacecast ptr addrspace(1) [[TMP6]] to ptr, !dbg [[DBG236]]
// CHECK1-NEXT: store ptr [[TMP7]], ptr [[_TMP2]], align 8, !dbg [[DBG236]] // CHECK1-NEXT: store ptr [[TMP7]], ptr [[_TMP2]], align 8, !dbg [[DBG236]]
// CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[_TMP2]], align 8, !dbg [[DBG236]] // CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[_TMP2]], align 8, !dbg [[DBG236]]
// CHECK1-NEXT: [[TMP9:%.*]] = load ptr addrspace(1), ptr [[BB_ADDR]], align 8, !dbg [[DBG236]] // CHECK1-NEXT: [[TMP9:%.*]] = load ptr addrspace(1), ptr [[BB_ADDR]], align 8, !dbg [[DBG236]]
// CHECK1-NEXT: [[TMP10:%.*]] = addrspacecast ptr addrspace(1) [[TMP9]] to ptr, !dbg [[DBG236]] // CHECK1-NEXT: [[TMP10:%.*]] = addrspacecast ptr addrspace(1) [[TMP9]] to ptr, !dbg [[DBG236]]
// CHECK1-NEXT: store ptr [[TMP10]], ptr [[_TMP3]], align 8, !dbg [[DBG236]] // CHECK1-NEXT: store ptr [[TMP10]], ptr [[_TMP3]], align 8, !dbg [[DBG236]]
// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[_TMP3]], align 8, !dbg [[DBG236]] // CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[_TMP3]], align 8, !dbg [[DBG236]]
// CHECK1-NEXT: [[TMP12:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB19:[0-9]+]], i8 2, i1 false, i1 true), !dbg [[DBG236]] // CHECK1-NEXT: [[TMP12:%.*]] = call i32 @__kmpc_target_init(ptr @[[GLOB19:[0-9]+]], i8 2, i1 false), !dbg [[DBG236]]
// CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP12]], -1, !dbg [[DBG236]] // CHECK1-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP12]], -1, !dbg [[DBG236]]
// CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]], !dbg [[DBG236]] // CHECK1-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]], !dbg [[DBG236]]
// CHECK1: user_code.entry: // CHECK1: user_code.entry:
// CHECK1-NEXT: [[TMP13:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB24:[0-9]+]]) // CHECK1-NEXT: [[TMP13:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB24:[0-9]+]])
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0, !dbg [[DBG237:![0-9]+]] // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 0, !dbg [[DBG237:![0-9]+]]
// CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP14]], align 8, !dbg [[DBG237]] // CHECK1-NEXT: store ptr [[TMP2]], ptr [[TMP14]], align 8, !dbg [[DBG237]]
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1, !dbg [[DBG237]] // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 1, !dbg [[DBG237]]
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP16]], align 8, !dbg [[DBG237]] // CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP15]], align 8, !dbg [[DBG237]]
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2, !dbg [[DBG237]] // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 2, !dbg [[DBG237]]
// CHECK1-NEXT: store ptr [[TMP8]], ptr [[TMP18]], align 8, !dbg [[DBG237]] // CHECK1-NEXT: store ptr [[TMP8]], ptr [[TMP16]], align 8, !dbg [[DBG237]]
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3, !dbg [[DBG237]] // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[CAPTURED_VARS_ADDRS]], i64 0, i64 3, !dbg [[DBG237]]
// CHECK1-NEXT: store ptr [[TMP11]], ptr [[TMP20]], align 8, !dbg [[DBG237]] // CHECK1-NEXT: store ptr [[TMP11]], ptr [[TMP17]], align 8, !dbg [[DBG237]]
// CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB24]], i32 [[TMP13]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__4, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4), !dbg [[DBG237]] // CHECK1-NEXT: call void @__kmpc_parallel_51(ptr @[[GLOB24]], i32 [[TMP13]], i32 1, i32 -1, i32 -1, ptr @__omp_outlined__4, ptr null, ptr [[CAPTURED_VARS_ADDRS]], i64 4), !dbg [[DBG237]]
// CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB26:[0-9]+]], i8 2, i1 true), !dbg [[DBG238:![0-9]+]] // CHECK1-NEXT: call void @__kmpc_target_deinit(ptr @[[GLOB26:[0-9]+]], i8 2), !dbg [[DBG238:![0-9]+]]
// CHECK1-NEXT: ret void, !dbg [[DBG240:![0-9]+]] // CHECK1-NEXT: ret void, !dbg [[DBG240:![0-9]+]]
// CHECK1: worker.exit: // CHECK1: worker.exit:
// CHECK1-NEXT: ret void, !dbg [[DBG236]] // CHECK1-NEXT: ret void, !dbg [[DBG236]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@__omp_outlined___debug__3 // CHECK1-LABEL: define {{[^@]+}}@__omp_outlined___debug__3
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr addrspace(1) noalias noundef [[C:%.*]], ptr addrspace(1) noalias noundef [[A:%.*]], ptr addrspace(1) noalias noundef [[B:%.*]], ptr addrspace(1) noalias noundef [[BB:%.*]]) #[[ATTR0]] !dbg [[DBG241:![0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr addrspace(1) noalias noundef [[C:%.*]], ptr addrspace(1) noalias noundef [[A:%.*]], ptr addrspace(1) noalias noundef [[B:%.*]], ptr addrspace(1) noalias noundef [[BB:%.*]]) #[[ATTR0]] !dbg [[DBG241:![0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 231 Lines • Show Last 20 Lines

clang/test/OpenMP/target_parallel_for_reduction_task_codegen.cpp

Show First 20 Lines • Show All 132 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP19:%.*]] = sub i64 [[TMP17]], [[TMP18]] // CHECK1-NEXT: [[TMP19:%.*]] = sub i64 [[TMP17]], [[TMP18]]
// CHECK1-NEXT: [[TMP20:%.*]] = sdiv exact i64 [[TMP19]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP20:%.*]] = sdiv exact i64 [[TMP19]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[VLA]], i64 [[TMP20]] // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[VLA]], i64 [[TMP20]]
// CHECK1-NEXT: store ptr [[_TMP6]], ptr [[_TMP5]], align 8 // CHECK1-NEXT: store ptr [[_TMP6]], ptr [[_TMP5]], align 8
// CHECK1-NEXT: store ptr [[TMP21]], ptr [[_TMP6]], align 8 // CHECK1-NEXT: store ptr [[TMP21]], ptr [[_TMP6]], align 8
// CHECK1-NEXT: [[DOTRD_INPUT_GEP_:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], ptr [[DOTRD_INPUT_]], i64 0, i64 0 // CHECK1-NEXT: [[DOTRD_INPUT_GEP_:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], ptr [[DOTRD_INPUT_]], i64 0, i64 0
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T:%.*]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 0 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T:%.*]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP22]], align 8 // CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP22]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 1 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP24]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP23]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 2 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 2
// CHECK1-NEXT: store i64 4, ptr [[TMP26]], align 8 // CHECK1-NEXT: store i64 4, ptr [[TMP24]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 3 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 3
// CHECK1-NEXT: store ptr @.red_init., ptr [[TMP27]], align 8 // CHECK1-NEXT: store ptr @.red_init., ptr [[TMP25]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 4 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 4
// CHECK1-NEXT: store ptr null, ptr [[TMP28]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP26]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 5 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.red_comb., ptr [[TMP29]], align 8 // CHECK1-NEXT: store ptr @.red_comb., ptr [[TMP27]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 6 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 6
// CHECK1-NEXT: call void @llvm.memset.p0.i64(ptr align 8 [[TMP30]], i8 0, i64 4, i1 false) // CHECK1-NEXT: call void @llvm.memset.p0.i64(ptr align 8 [[TMP28]], i8 0, i64 4, i1 false)
// CHECK1-NEXT: [[DOTRD_INPUT_GEP_7:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], ptr [[DOTRD_INPUT_]], i64 0, i64 1 // CHECK1-NEXT: [[DOTRD_INPUT_GEP_7:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], ptr [[DOTRD_INPUT_]], i64 0, i64 1
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 0 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 0
// CHECK1-NEXT: [[TMP33:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8 // CHECK1-NEXT: [[TMP30:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8
// CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds ptr, ptr [[TMP33]], i64 0 // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds ptr, ptr [[TMP30]], i64 0
// CHECK1-NEXT: [[TMP34:%.*]] = load ptr, ptr [[ARRAYIDX8]], align 8 // CHECK1-NEXT: [[TMP31:%.*]] = load ptr, ptr [[ARRAYIDX8]], align 8
// CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds i8, ptr [[TMP34]], i64 0 // CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds i8, ptr [[TMP31]], i64 0
// CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK1-NEXT: [[TMP33:%.*]] = sext i32 [[TMP32]] to i64
// CHECK1-NEXT: [[LB_ADD_LEN10:%.*]] = add nsw i64 -1, [[TMP36]] // CHECK1-NEXT: [[LB_ADD_LEN10:%.*]] = add nsw i64 -1, [[TMP33]]
// CHECK1-NEXT: [[TMP37:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8 // CHECK1-NEXT: [[TMP34:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8
// CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds ptr, ptr [[TMP37]], i64 9 // CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds ptr, ptr [[TMP34]], i64 9
// CHECK1-NEXT: [[TMP38:%.*]] = load ptr, ptr [[ARRAYIDX11]], align 8 // CHECK1-NEXT: [[TMP35:%.*]] = load ptr, ptr [[ARRAYIDX11]], align 8
// CHECK1-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds i8, ptr [[TMP38]], i64 [[LB_ADD_LEN10]] // CHECK1-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds i8, ptr [[TMP35]], i64 [[LB_ADD_LEN10]]
// CHECK1-NEXT: store ptr [[VLA]], ptr [[TMP32]], align 8 // CHECK1-NEXT: store ptr [[VLA]], ptr [[TMP29]], align 8
// CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 1 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[ARRAYIDX9]], ptr [[TMP39]], align 8 // CHECK1-NEXT: store ptr [[ARRAYIDX9]], ptr [[TMP36]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = ptrtoint ptr [[ARRAYIDX12]] to i64 // CHECK1-NEXT: [[TMP37:%.*]] = ptrtoint ptr [[ARRAYIDX12]] to i64
// CHECK1-NEXT: [[TMP41:%.*]] = ptrtoint ptr [[ARRAYIDX9]] to i64 // CHECK1-NEXT: [[TMP38:%.*]] = ptrtoint ptr [[ARRAYIDX9]] to i64
// CHECK1-NEXT: [[TMP42:%.*]] = sub i64 [[TMP40]], [[TMP41]] // CHECK1-NEXT: [[TMP39:%.*]] = sub i64 [[TMP37]], [[TMP38]]
// CHECK1-NEXT: [[TMP43:%.*]] = sdiv exact i64 [[TMP42]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP40:%.*]] = sdiv exact i64 [[TMP39]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP44:%.*]] = add nuw i64 [[TMP43]], 1 // CHECK1-NEXT: [[TMP41:%.*]] = add nuw i64 [[TMP40]], 1
// CHECK1-NEXT: [[TMP45:%.*]] = mul nuw i64 [[TMP44]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP42:%.*]] = mul nuw i64 [[TMP41]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 2 // CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 2
// CHECK1-NEXT: store i64 [[TMP45]], ptr [[TMP46]], align 8 // CHECK1-NEXT: store i64 [[TMP42]], ptr [[TMP43]], align 8
// CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 3 // CHECK1-NEXT: [[TMP44:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 3
// CHECK1-NEXT: store ptr @.red_init..1, ptr [[TMP47]], align 8 // CHECK1-NEXT: store ptr @.red_init..1, ptr [[TMP44]], align 8
// CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 4 // CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 4
// CHECK1-NEXT: store ptr null, ptr [[TMP48]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP45]], align 8
// CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 5 // CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.red_comb..2, ptr [[TMP49]], align 8 // CHECK1-NEXT: store ptr @.red_comb..2, ptr [[TMP46]], align 8
// CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 6 // CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 6
// CHECK1-NEXT: store i32 1, ptr [[TMP50]], align 8 // CHECK1-NEXT: store i32 1, ptr [[TMP47]], align 8
// CHECK1-NEXT: [[TMP48:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP49:%.*]] = load i32, ptr [[TMP48]], align 4
// CHECK1-NEXT: [[TMP50:%.*]] = call ptr @__kmpc_taskred_modifier_init(ptr @[[GLOB1]], i32 [[TMP49]], i32 1, i32 2, ptr [[DOTRD_INPUT_]])
// CHECK1-NEXT: store ptr [[TMP50]], ptr [[DOTTASK_RED_]], align 8
// CHECK1-NEXT: [[TMP51:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP51:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP52:%.*]] = load i32, ptr [[TMP51]], align 4 // CHECK1-NEXT: [[TMP52:%.*]] = load i32, ptr [[TMP51]], align 4
// CHECK1-NEXT: [[TMP54:%.*]] = call ptr @__kmpc_taskred_modifier_init(ptr @[[GLOB1]], i32 [[TMP52]], i32 1, i32 2, ptr [[DOTRD_INPUT_]]) // CHECK1-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB2:[0-9]+]], i32 [[TMP52]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK1-NEXT: store ptr [[TMP54]], ptr [[DOTTASK_RED_]], align 8 // CHECK1-NEXT: [[TMP53:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK1-NEXT: [[TMP55:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i64 [[TMP53]], 9
// CHECK1-NEXT: [[TMP56:%.*]] = load i32, ptr [[TMP55]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB2:[0-9]+]], i32 [[TMP56]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK1-NEXT: [[TMP57:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i64 [[TMP57]], 9
// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1: cond.true: // CHECK1: cond.true:
// CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1-NEXT: br label [[COND_END:%.*]]
// CHECK1: cond.false: // CHECK1: cond.false:
// CHECK1-NEXT: [[TMP58:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8 // CHECK1-NEXT: [[TMP54:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK1-NEXT: br label [[COND_END]] // CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end: // CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i64 [ 9, [[COND_TRUE]] ], [ [[TMP58]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i64 [ 9, [[COND_TRUE]] ], [ [[TMP54]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i64 [[COND]], ptr [[DOTOMP_UB]], align 8 // CHECK1-NEXT: store i64 [[COND]], ptr [[DOTOMP_UB]], align 8
// CHECK1-NEXT: [[TMP59:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8 // CHECK1-NEXT: [[TMP55:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK1-NEXT: store i64 [[TMP59]], ptr [[DOTOMP_IV]], align 8 // CHECK1-NEXT: store i64 [[TMP55]], ptr [[DOTOMP_IV]], align 8
// 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: [[TMP60:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK1-NEXT: [[TMP56:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK1-NEXT: [[TMP61:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8 // CHECK1-NEXT: [[TMP57:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK1-NEXT: [[CMP13:%.*]] = icmp sle i64 [[TMP60]], [[TMP61]] // CHECK1-NEXT: [[CMP13:%.*]] = icmp sle i64 [[TMP56]], [[TMP57]]
// CHECK1-NEXT: br i1 [[CMP13]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]] // CHECK1-NEXT: br i1 [[CMP13]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1: omp.inner.for.cond.cleanup: // CHECK1: omp.inner.for.cond.cleanup:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_END:%.*]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body: // CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP62:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK1-NEXT: [[TMP58:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i64 [[TMP62]], 1 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i64 [[TMP58]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL]]
// CHECK1-NEXT: store i64 [[ADD]], ptr [[I]], align 8 // CHECK1-NEXT: store i64 [[ADD]], ptr [[I]], align 8
// CHECK1-NEXT: [[TMP63:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 0 // CHECK1-NEXT: [[TMP59:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[DOTTASK_RED_]], ptr [[TMP63]], align 8 // CHECK1-NEXT: store ptr [[DOTTASK_RED_]], ptr [[TMP59]], align 8
// CHECK1-NEXT: [[TMP64:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 1 // CHECK1-NEXT: [[TMP60:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP64]], align 8 // CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP60]], align 8
// CHECK1-NEXT: [[TMP65:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 2 // CHECK1-NEXT: [[TMP61:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 2
// CHECK1-NEXT: [[TMP66:%.*]] = load ptr, ptr [[_TMP5]], align 8 // CHECK1-NEXT: [[TMP62:%.*]] = load ptr, ptr [[_TMP5]], align 8
// CHECK1-NEXT: store ptr [[TMP66]], ptr [[TMP65]], align 8 // CHECK1-NEXT: store ptr [[TMP62]], ptr [[TMP61]], align 8
// CHECK1-NEXT: [[TMP67:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP63:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP68:%.*]] = load i32, ptr [[TMP67]], align 4 // CHECK1-NEXT: [[TMP64:%.*]] = load i32, ptr [[TMP63]], align 4
// CHECK1-NEXT: [[TMP69:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP68]], i32 1, i64 48, i64 24, ptr @.omp_task_entry.) // CHECK1-NEXT: [[TMP65:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP64]], i32 1, i64 48, i64 24, ptr @.omp_task_entry.)
// CHECK1-NEXT: [[TMP71:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP69]], i32 0, i32 0 // CHECK1-NEXT: [[TMP66:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP65]], i32 0, i32 0
// CHECK1-NEXT: [[TMP72:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP71]], i32 0, i32 0 // CHECK1-NEXT: [[TMP67:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP66]], i32 0, i32 0
// CHECK1-NEXT: [[TMP73:%.*]] = load ptr, ptr [[TMP72]], align 8 // CHECK1-NEXT: [[TMP68:%.*]] = load ptr, ptr [[TMP67]], align 8
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP73]], ptr align 8 [[AGG_CAPTURED]], i64 24, i1 false) // CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP68]], ptr align 8 [[AGG_CAPTURED]], i64 24, i1 false)
// CHECK1-NEXT: [[TMP75:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP69]], i32 0, i32 1 // CHECK1-NEXT: [[TMP69:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP65]], i32 0, i32 1
// CHECK1-NEXT: [[TMP76:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP75]], i32 0, i32 0 // CHECK1-NEXT: [[TMP70:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP69]], i32 0, i32 0
// CHECK1-NEXT: [[TMP77:%.*]] = load ptr, ptr [[DOTTASK_RED_]], align 8 // CHECK1-NEXT: [[TMP71:%.*]] = load ptr, ptr [[DOTTASK_RED_]], align 8
// CHECK1-NEXT: store ptr [[TMP77]], ptr [[TMP76]], align 8 // CHECK1-NEXT: store ptr [[TMP71]], ptr [[TMP70]], align 8
// CHECK1-NEXT: [[TMP78:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP72:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP79:%.*]] = load i32, ptr [[TMP78]], align 4 // CHECK1-NEXT: [[TMP73:%.*]] = load i32, ptr [[TMP72]], align 4
// CHECK1-NEXT: [[TMP80:%.*]] = call i32 @__kmpc_omp_task(ptr @[[GLOB1]], i32 [[TMP79]], ptr [[TMP69]]) // CHECK1-NEXT: [[TMP74:%.*]] = call i32 @__kmpc_omp_task(ptr @[[GLOB1]], i32 [[TMP73]], ptr [[TMP65]])
// 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: [[TMP81:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK1-NEXT: [[TMP75:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK1-NEXT: [[ADD14:%.*]] = add nsw i64 [[TMP81]], 1 // CHECK1-NEXT: [[ADD14:%.*]] = add nsw i64 [[TMP75]], 1
// CHECK1-NEXT: store i64 [[ADD14]], ptr [[DOTOMP_IV]], align 8 // CHECK1-NEXT: store i64 [[ADD14]], ptr [[DOTOMP_IV]], align 8
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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: [[TMP82:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP76:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP83:%.*]] = load i32, ptr [[TMP82]], align 4 // CHECK1-NEXT: [[TMP77:%.*]] = load i32, ptr [[TMP76]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP83]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP77]])
// CHECK1-NEXT: [[TMP78:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP79:%.*]] = load i32, ptr [[TMP78]], align 4
// CHECK1-NEXT: call void @__kmpc_task_reduction_modifier_fini(ptr @[[GLOB1]], i32 [[TMP79]], i32 1)
// CHECK1-NEXT: [[TMP80:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP80]], align 8
// CHECK1-NEXT: [[TMP81:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT: store ptr [[VLA]], ptr [[TMP81]], align 8
// CHECK1-NEXT: [[TMP82:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK1-NEXT: [[TMP83:%.*]] = inttoptr i64 [[TMP11]] to ptr
// CHECK1-NEXT: store ptr [[TMP83]], ptr [[TMP82]], align 8
// CHECK1-NEXT: [[TMP84:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP84:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP85:%.*]] = load i32, ptr [[TMP84]], align 4 // CHECK1-NEXT: [[TMP85:%.*]] = load i32, ptr [[TMP84]], align 4
// CHECK1-NEXT: call void @__kmpc_task_reduction_modifier_fini(ptr @[[GLOB1]], i32 [[TMP85]], i32 1) // CHECK1-NEXT: [[TMP86:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3:[0-9]+]], i32 [[TMP85]], i32 2, i64 24, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: [[TMP86:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: switch i32 [[TMP86]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP86]], align 8
// CHECK1-NEXT: [[TMP88:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT: store ptr [[VLA]], ptr [[TMP88]], align 8
// CHECK1-NEXT: [[TMP89:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK1-NEXT: [[TMP90:%.*]] = inttoptr i64 [[TMP11]] to ptr
// CHECK1-NEXT: store ptr [[TMP90]], ptr [[TMP89]], align 8
// CHECK1-NEXT: [[TMP91:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP92:%.*]] = load i32, ptr [[TMP91]], align 4
// CHECK1-NEXT: [[TMP94:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3:[0-9]+]], i32 [[TMP92]], i32 2, i64 24, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP94]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP95:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP87:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP96:%.*]] = load i32, ptr [[ARGC1]], align 4 // CHECK1-NEXT: [[TMP88:%.*]] = load i32, ptr [[ARGC1]], align 4
// CHECK1-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP95]], [[TMP96]] // CHECK1-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP87]], [[TMP88]]
// CHECK1-NEXT: store i32 [[ADD15]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD15]], ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP97:%.*]] = getelementptr i8, ptr [[ARRAYIDX2]], i64 [[TMP11]] // CHECK1-NEXT: [[TMP89:%.*]] = getelementptr i8, ptr [[ARRAYIDX2]], i64 [[TMP11]]
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX2]], [[TMP97]] // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX2]], [[TMP89]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE22:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE22:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1: omp.arraycpy.body: // CHECK1: omp.arraycpy.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST16:%.*]] = phi ptr [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT20:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST16:%.*]] = phi ptr [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT20:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[TMP98:%.*]] = load i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1 // CHECK1-NEXT: [[TMP90:%.*]] = load i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1
// CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP98]] to i32 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP90]] to i32
// CHECK1-NEXT: [[TMP99:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 // CHECK1-NEXT: [[TMP91:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
// CHECK1-NEXT: [[CONV17:%.*]] = sext i8 [[TMP99]] to i32 // CHECK1-NEXT: [[CONV17:%.*]] = sext i8 [[TMP91]] to i32
// CHECK1-NEXT: [[ADD18:%.*]] = add nsw i32 [[CONV]], [[CONV17]] // CHECK1-NEXT: [[ADD18:%.*]] = add nsw i32 [[CONV]], [[CONV17]]
// CHECK1-NEXT: [[CONV19:%.*]] = trunc i32 [[ADD18]] to i8 // CHECK1-NEXT: [[CONV19:%.*]] = trunc i32 [[ADD18]] to i8
// CHECK1-NEXT: store i8 [[CONV19]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1 // CHECK1-NEXT: store i8 [[CONV19]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT20]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST16]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT20]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST16]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE21:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT20]], [[TMP97]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE21:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT20]], [[TMP89]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_DONE22]], label [[OMP_ARRAYCPY_BODY]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_DONE22]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1: omp.arraycpy.done22: // CHECK1: omp.arraycpy.done22:
// CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP92]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP85]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP100:%.*]] = load i32, ptr [[ARGC1]], align 4 // CHECK1-NEXT: [[TMP92:%.*]] = load i32, ptr [[ARGC1]], align 4
// CHECK1-NEXT: [[TMP101:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP100]] monotonic, align 4 // CHECK1-NEXT: [[TMP93:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP92]] monotonic, align 4
// CHECK1-NEXT: [[TMP102:%.*]] = getelementptr i8, ptr [[ARRAYIDX2]], i64 [[TMP11]] // CHECK1-NEXT: [[TMP94:%.*]] = getelementptr i8, ptr [[ARRAYIDX2]], i64 [[TMP11]]
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY23:%.*]] = icmp eq ptr [[ARRAYIDX2]], [[TMP102]] // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY23:%.*]] = icmp eq ptr [[ARRAYIDX2]], [[TMP94]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY23]], label [[OMP_ARRAYCPY_DONE36:%.*]], label [[OMP_ARRAYCPY_BODY24:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY23]], label [[OMP_ARRAYCPY_DONE36:%.*]], label [[OMP_ARRAYCPY_BODY24:%.*]]
// CHECK1: omp.arraycpy.body24: // CHECK1: omp.arraycpy.body24:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST25:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT34:%.*]], [[ATOMIC_EXIT:%.*]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST25:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT34:%.*]], [[ATOMIC_EXIT:%.*]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST26:%.*]] = phi ptr [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT33:%.*]], [[ATOMIC_EXIT]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST26:%.*]] = phi ptr [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT33:%.*]], [[ATOMIC_EXIT]] ]
// CHECK1-NEXT: [[TMP103:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1 // CHECK1-NEXT: [[TMP95:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1
// CHECK1-NEXT: [[CONV27:%.*]] = sext i8 [[TMP103]] to i32 // CHECK1-NEXT: [[CONV27:%.*]] = sext i8 [[TMP95]] to i32
// CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST26]] monotonic, align 1 // CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST26]] monotonic, align 1
// CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]] // CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]]
// CHECK1: atomic_cont: // CHECK1: atomic_cont:
// CHECK1-NEXT: [[TMP104:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY24]] ], [ [[TMP109:%.*]], [[ATOMIC_CONT]] ] // CHECK1-NEXT: [[TMP96:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY24]] ], [ [[TMP101:%.*]], [[ATOMIC_CONT]] ]
// CHECK1-NEXT: store i8 [[TMP104]], ptr [[_TMP28]], align 1 // CHECK1-NEXT: store i8 [[TMP96]], ptr [[_TMP28]], align 1
// CHECK1-NEXT: [[TMP105:%.*]] = load i8, ptr [[_TMP28]], align 1 // CHECK1-NEXT: [[TMP97:%.*]] = load i8, ptr [[_TMP28]], align 1
// CHECK1-NEXT: [[CONV29:%.*]] = sext i8 [[TMP105]] to i32 // CHECK1-NEXT: [[CONV29:%.*]] = sext i8 [[TMP97]] to i32
// CHECK1-NEXT: [[TMP106:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1 // CHECK1-NEXT: [[TMP98:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1
// CHECK1-NEXT: [[CONV30:%.*]] = sext i8 [[TMP106]] to i32 // CHECK1-NEXT: [[CONV30:%.*]] = sext i8 [[TMP98]] to i32
// CHECK1-NEXT: [[ADD31:%.*]] = add nsw i32 [[CONV29]], [[CONV30]] // CHECK1-NEXT: [[ADD31:%.*]] = add nsw i32 [[CONV29]], [[CONV30]]
// CHECK1-NEXT: [[CONV32:%.*]] = trunc i32 [[ADD31]] to i8 // CHECK1-NEXT: [[CONV32:%.*]] = trunc i32 [[ADD31]] to i8
// CHECK1-NEXT: store i8 [[CONV32]], ptr [[ATOMIC_TEMP]], align 1 // CHECK1-NEXT: store i8 [[CONV32]], ptr [[ATOMIC_TEMP]], align 1
// CHECK1-NEXT: [[TMP107:%.*]] = load i8, ptr [[ATOMIC_TEMP]], align 1 // CHECK1-NEXT: [[TMP99:%.*]] = load i8, ptr [[ATOMIC_TEMP]], align 1
// CHECK1-NEXT: [[TMP108:%.*]] = cmpxchg ptr [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i8 [[TMP104]], i8 [[TMP107]] monotonic monotonic, align 1 // CHECK1-NEXT: [[TMP100:%.*]] = cmpxchg ptr [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i8 [[TMP96]], i8 [[TMP99]] monotonic monotonic, align 1
// CHECK1-NEXT: [[TMP109]] = extractvalue { i8, i1 } [[TMP108]], 0 // CHECK1-NEXT: [[TMP101]] = extractvalue { i8, i1 } [[TMP100]], 0
// CHECK1-NEXT: [[TMP110:%.*]] = extractvalue { i8, i1 } [[TMP108]], 1 // CHECK1-NEXT: [[TMP102:%.*]] = extractvalue { i8, i1 } [[TMP100]], 1
// CHECK1-NEXT: br i1 [[TMP110]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]] // CHECK1-NEXT: br i1 [[TMP102]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]]
// CHECK1: atomic_exit: // CHECK1: atomic_exit:
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT33]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT33]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT34]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST25]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT34]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST25]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE35:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT33]], [[TMP102]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE35:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT33]], [[TMP94]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE35]], label [[OMP_ARRAYCPY_DONE36]], label [[OMP_ARRAYCPY_BODY24]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE35]], label [[OMP_ARRAYCPY_DONE36]], label [[OMP_ARRAYCPY_BODY24]]
// CHECK1: omp.arraycpy.done36: // CHECK1: omp.arraycpy.done36:
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: [[TMP111:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 // CHECK1-NEXT: [[TMP103:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK1-NEXT: call void @llvm.stackrestore(ptr [[TMP111]]) // CHECK1-NEXT: call void @llvm.stackrestore(ptr [[TMP103]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.red_init. // CHECK1-LABEL: define {{[^@]+}}@.red_init.
// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store i32 0, ptr [[TMP3]], align 4 // CHECK1-NEXT: store i32 0, ptr [[TMP2]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.red_comb. // CHECK1-LABEL: define {{[^@]+}}@.red_comb.
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP3]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP5]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP6]], [[TMP7]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP4]], [[TMP5]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP3]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP2]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.red_init..1 // CHECK1-LABEL: define {{[^@]+}}@.red_init..1
// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
▲ Show 20 Lines • Show All 78 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP3]], i32 0, i32 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP3]], i32 0, i32 0
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP3]], i32 0, i32 1 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP3]], i32 0, i32 1
// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META3:![0-9]+]]) // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META3:![0-9]+]])
// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META6:![0-9]+]]) // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META6:![0-9]+]])
// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META8:![0-9]+]]) // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META8:![0-9]+]])
// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]]) // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]])
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: store ptr [[TMP9]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: store ptr @.omp_task_privates_map., ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr @.omp_task_privates_map., ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: call void [[TMP13]](ptr [[TMP14]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR6]] // CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR6]]
// CHECK1-NEXT: [[TMP16:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], ptr [[TMP12]], i32 0, i32 1 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], ptr [[TMP9]], i32 0, i32 1
// CHECK1-NEXT: [[TMP18:%.*]] = load ptr, ptr [[TMP17]], align 8 // CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = load ptr, ptr [[TMP16]], align 8 // CHECK1-NEXT: [[TMP15:%.*]] = load ptr, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12
// CHECK1-NEXT: [[TMP22:%.*]] = call ptr @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], ptr [[TMP19]], ptr [[TMP18]]) // CHECK1-NEXT: [[TMP17:%.*]] = call ptr @__kmpc_task_reduction_get_th_data(i32 [[TMP16]], ptr [[TMP15]], ptr [[TMP14]])
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP12]], i32 0, i32 2 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP9]], i32 0, i32 2
// CHECK1-NEXT: [[TMP24:%.*]] = load ptr, ptr [[TMP23]], align 8 // CHECK1-NEXT: [[TMP19:%.*]] = load ptr, ptr [[TMP18]], align 8
// CHECK1-NEXT: [[TMP25:%.*]] = load ptr, ptr [[TMP24]], align 8 // CHECK1-NEXT: [[TMP20:%.*]] = load ptr, ptr [[TMP19]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP12]], i32 0, i32 1 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP9]], i32 0, i32 1
// CHECK1-NEXT: [[TMP27:%.*]] = load ptr, ptr [[TMP26]], align 8 // CHECK1-NEXT: [[TMP22:%.*]] = load ptr, ptr [[TMP21]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4 // CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
// CHECK1-NEXT: [[TMP29:%.*]] = sext i32 [[TMP28]] to i64 // CHECK1-NEXT: [[TMP24:%.*]] = sext i32 [[TMP23]] to i64
// CHECK1-NEXT: [[LB_ADD_LEN_I:%.*]] = add nsw i64 -1, [[TMP29]] // CHECK1-NEXT: [[LB_ADD_LEN_I:%.*]] = add nsw i64 -1, [[TMP24]]
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP12]], i32 0, i32 2 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP9]], i32 0, i32 2
// CHECK1-NEXT: [[TMP31:%.*]] = load ptr, ptr [[TMP30]], align 8 // CHECK1-NEXT: [[TMP26:%.*]] = load ptr, ptr [[TMP25]], align 8
// CHECK1-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds ptr, ptr [[TMP31]], i64 9 // CHECK1-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds ptr, ptr [[TMP26]], i64 9
// CHECK1-NEXT: [[TMP32:%.*]] = load ptr, ptr [[ARRAYIDX2_I]], align 8 // CHECK1-NEXT: [[TMP27:%.*]] = load ptr, ptr [[ARRAYIDX2_I]], align 8
// CHECK1-NEXT: [[ARRAYIDX3_I:%.*]] = getelementptr inbounds i8, ptr [[TMP32]], i64 [[LB_ADD_LEN_I]] // CHECK1-NEXT: [[ARRAYIDX3_I:%.*]] = getelementptr inbounds i8, ptr [[TMP27]], i64 [[LB_ADD_LEN_I]]
// CHECK1-NEXT: [[TMP33:%.*]] = ptrtoint ptr [[ARRAYIDX3_I]] to i64 // CHECK1-NEXT: [[TMP28:%.*]] = ptrtoint ptr [[ARRAYIDX3_I]] to i64
// CHECK1-NEXT: [[TMP34:%.*]] = ptrtoint ptr [[TMP25]] to i64 // CHECK1-NEXT: [[TMP29:%.*]] = ptrtoint ptr [[TMP20]] to i64
// CHECK1-NEXT: [[TMP35:%.*]] = sub i64 [[TMP33]], [[TMP34]] // CHECK1-NEXT: [[TMP30:%.*]] = sub i64 [[TMP28]], [[TMP29]]
// CHECK1-NEXT: [[TMP36:%.*]] = sdiv exact i64 [[TMP35]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP31:%.*]] = sdiv exact i64 [[TMP30]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP37:%.*]] = add nuw i64 [[TMP36]], 1 // CHECK1-NEXT: [[TMP32:%.*]] = add nuw i64 [[TMP31]], 1
// CHECK1-NEXT: [[TMP38:%.*]] = mul nuw i64 [[TMP37]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP33:%.*]] = mul nuw i64 [[TMP32]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: store i64 [[TMP37]], ptr @{{reduction_size[.].+[.]}}, align 8, !noalias !12 // CHECK1-NEXT: store i64 [[TMP32]], ptr @{{reduction_size[.].+[.]}}, align 8, !noalias !12
// CHECK1-NEXT: [[TMP39:%.*]] = load ptr, ptr [[TMP16]], align 8 // CHECK1-NEXT: [[TMP34:%.*]] = load ptr, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = call ptr @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], ptr [[TMP39]], ptr [[TMP25]]) // CHECK1-NEXT: [[TMP35:%.*]] = call ptr @__kmpc_task_reduction_get_th_data(i32 [[TMP16]], ptr [[TMP34]], ptr [[TMP20]])
// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP12]], i32 0, i32 2 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP9]], i32 0, i32 2
// CHECK1-NEXT: [[TMP42:%.*]] = load ptr, ptr [[TMP41]], align 8 // CHECK1-NEXT: [[TMP37:%.*]] = load ptr, ptr [[TMP36]], align 8
// CHECK1-NEXT: [[TMP43:%.*]] = load ptr, ptr [[TMP42]], align 8 // CHECK1-NEXT: [[TMP38:%.*]] = load ptr, ptr [[TMP37]], align 8
// CHECK1-NEXT: [[TMP44:%.*]] = ptrtoint ptr [[TMP43]] to i64 // CHECK1-NEXT: [[TMP39:%.*]] = ptrtoint ptr [[TMP38]] to i64
// CHECK1-NEXT: [[TMP45:%.*]] = ptrtoint ptr [[TMP25]] to i64 // CHECK1-NEXT: [[TMP40:%.*]] = ptrtoint ptr [[TMP20]] to i64
// CHECK1-NEXT: [[TMP46:%.*]] = sub i64 [[TMP44]], [[TMP45]] // CHECK1-NEXT: [[TMP41:%.*]] = sub i64 [[TMP39]], [[TMP40]]
// CHECK1-NEXT: [[TMP47:%.*]] = sdiv exact i64 [[TMP46]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP42:%.*]] = sdiv exact i64 [[TMP41]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP48:%.*]] = getelementptr i8, ptr [[TMP40]], i64 [[TMP47]] // CHECK1-NEXT: [[TMP43:%.*]] = getelementptr i8, ptr [[TMP35]], i64 [[TMP42]]
// CHECK1-NEXT: store ptr [[TMP4_I]], ptr [[TMP_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP4_I]], ptr [[TMP_I]], align 8, !noalias !12
// CHECK1-NEXT: store ptr [[TMP48]], ptr [[TMP4_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP43]], ptr [[TMP4_I]], align 8, !noalias !12
// CHECK1-NEXT: ret i32 0 // CHECK1-NEXT: ret i32 0
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP3]], i64 0, i64 1
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP4]], i64 0, i64 1 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 1
// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 2
// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[TMP12]], align 8 // CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 1 // CHECK1-NEXT: [[TMP14:%.*]] = ptrtoint ptr [[TMP13]] to i64
// CHECK1-NEXT: [[TMP15:%.*]] = load ptr, ptr [[TMP14]], align 8 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 2 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP17:%.*]] = load ptr, ptr [[TMP16]], align 8 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP15]], [[TMP16]]
// CHECK1-NEXT: [[TMP18:%.*]] = ptrtoint ptr [[TMP17]] to i64 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[TMP11]], i64 [[TMP14]]
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP11]], [[TMP17]]
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[TMP15]], i64 [[TMP18]]
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP15]], [[TMP21]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1: omp.arraycpy.body: // CHECK1: omp.arraycpy.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP13]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP9]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP15]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[TMP22:%.*]] = load i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 // CHECK1-NEXT: [[TMP18:%.*]] = load i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP22]] to i32 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP18]] to i32
// CHECK1-NEXT: [[TMP23:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 // CHECK1-NEXT: [[TMP19:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
// CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP23]] to i32 // CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP19]] to i32
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]] // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]]
// CHECK1-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8 // CHECK1-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8
// CHECK1-NEXT: store i8 [[CONV4]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 // CHECK1-NEXT: store i8 [[CONV4]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP21]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP17]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1: omp.arraycpy.done5: // CHECK1: omp.arraycpy.done5:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //

clang/test/OpenMP/target_parallel_reduction_task_codegen.cpp

Show First 20 Lines • Show All 121 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP19:%.*]] = sub i64 [[TMP17]], [[TMP18]] // CHECK1-NEXT: [[TMP19:%.*]] = sub i64 [[TMP17]], [[TMP18]]
// CHECK1-NEXT: [[TMP20:%.*]] = sdiv exact i64 [[TMP19]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP20:%.*]] = sdiv exact i64 [[TMP19]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[VLA]], i64 [[TMP20]] // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[VLA]], i64 [[TMP20]]
// CHECK1-NEXT: store ptr [[_TMP5]], ptr [[TMP]], align 8 // CHECK1-NEXT: store ptr [[_TMP5]], ptr [[TMP]], align 8
// CHECK1-NEXT: store ptr [[TMP21]], ptr [[_TMP5]], align 8 // CHECK1-NEXT: store ptr [[TMP21]], ptr [[_TMP5]], align 8
// CHECK1-NEXT: [[DOTRD_INPUT_GEP_:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], ptr [[DOTRD_INPUT_]], i64 0, i64 0 // CHECK1-NEXT: [[DOTRD_INPUT_GEP_:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], ptr [[DOTRD_INPUT_]], i64 0, i64 0
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T:%.*]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 0 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T:%.*]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP22]], align 8 // CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP22]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 1 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP24]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP23]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 2 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 2
// CHECK1-NEXT: store i64 4, ptr [[TMP26]], align 8 // CHECK1-NEXT: store i64 4, ptr [[TMP24]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 3 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 3
// CHECK1-NEXT: store ptr @.red_init., ptr [[TMP27]], align 8 // CHECK1-NEXT: store ptr @.red_init., ptr [[TMP25]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 4 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 4
// CHECK1-NEXT: store ptr null, ptr [[TMP28]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP26]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 5 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.red_comb., ptr [[TMP29]], align 8 // CHECK1-NEXT: store ptr @.red_comb., ptr [[TMP27]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 6 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 6
// CHECK1-NEXT: call void @llvm.memset.p0.i64(ptr align 8 [[TMP30]], i8 0, i64 4, i1 false) // CHECK1-NEXT: call void @llvm.memset.p0.i64(ptr align 8 [[TMP28]], i8 0, i64 4, i1 false)
// CHECK1-NEXT: [[DOTRD_INPUT_GEP_6:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], ptr [[DOTRD_INPUT_]], i64 0, i64 1 // CHECK1-NEXT: [[DOTRD_INPUT_GEP_6:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], ptr [[DOTRD_INPUT_]], i64 0, i64 1
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 0 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 0
// CHECK1-NEXT: [[TMP33:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8 // CHECK1-NEXT: [[TMP30:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8
// CHECK1-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds ptr, ptr [[TMP33]], i64 0 // CHECK1-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds ptr, ptr [[TMP30]], i64 0
// CHECK1-NEXT: [[TMP34:%.*]] = load ptr, ptr [[ARRAYIDX7]], align 8 // CHECK1-NEXT: [[TMP31:%.*]] = load ptr, ptr [[ARRAYIDX7]], align 8
// CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i8, ptr [[TMP34]], i64 0 // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i8, ptr [[TMP31]], i64 0
// CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK1-NEXT: [[TMP33:%.*]] = sext i32 [[TMP32]] to i64
// CHECK1-NEXT: [[LB_ADD_LEN9:%.*]] = add nsw i64 -1, [[TMP36]] // CHECK1-NEXT: [[LB_ADD_LEN9:%.*]] = add nsw i64 -1, [[TMP33]]
// CHECK1-NEXT: [[TMP37:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8 // CHECK1-NEXT: [[TMP34:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8
// CHECK1-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds ptr, ptr [[TMP37]], i64 9 // CHECK1-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds ptr, ptr [[TMP34]], i64 9
// CHECK1-NEXT: [[TMP38:%.*]] = load ptr, ptr [[ARRAYIDX10]], align 8 // CHECK1-NEXT: [[TMP35:%.*]] = load ptr, ptr [[ARRAYIDX10]], align 8
// CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds i8, ptr [[TMP38]], i64 [[LB_ADD_LEN9]] // CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds i8, ptr [[TMP35]], i64 [[LB_ADD_LEN9]]
// CHECK1-NEXT: store ptr [[VLA]], ptr [[TMP32]], align 8 // CHECK1-NEXT: store ptr [[VLA]], ptr [[TMP29]], align 8
// CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 1 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[ARRAYIDX8]], ptr [[TMP39]], align 8 // CHECK1-NEXT: store ptr [[ARRAYIDX8]], ptr [[TMP36]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = ptrtoint ptr [[ARRAYIDX11]] to i64 // CHECK1-NEXT: [[TMP37:%.*]] = ptrtoint ptr [[ARRAYIDX11]] to i64
// CHECK1-NEXT: [[TMP41:%.*]] = ptrtoint ptr [[ARRAYIDX8]] to i64 // CHECK1-NEXT: [[TMP38:%.*]] = ptrtoint ptr [[ARRAYIDX8]] to i64
// CHECK1-NEXT: [[TMP42:%.*]] = sub i64 [[TMP40]], [[TMP41]] // CHECK1-NEXT: [[TMP39:%.*]] = sub i64 [[TMP37]], [[TMP38]]
// CHECK1-NEXT: [[TMP43:%.*]] = sdiv exact i64 [[TMP42]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP40:%.*]] = sdiv exact i64 [[TMP39]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP44:%.*]] = add nuw i64 [[TMP43]], 1 // CHECK1-NEXT: [[TMP41:%.*]] = add nuw i64 [[TMP40]], 1
// CHECK1-NEXT: [[TMP45:%.*]] = mul nuw i64 [[TMP44]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP42:%.*]] = mul nuw i64 [[TMP41]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 2 // CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 2
// CHECK1-NEXT: store i64 [[TMP45]], ptr [[TMP46]], align 8 // CHECK1-NEXT: store i64 [[TMP42]], ptr [[TMP43]], align 8
// CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 3 // CHECK1-NEXT: [[TMP44:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 3
// CHECK1-NEXT: store ptr @.red_init..1, ptr [[TMP47]], align 8 // CHECK1-NEXT: store ptr @.red_init..1, ptr [[TMP44]], align 8
// CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 4 // CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 4
// CHECK1-NEXT: store ptr null, ptr [[TMP48]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP45]], align 8
// CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 5 // CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.red_comb..2, ptr [[TMP49]], align 8 // CHECK1-NEXT: store ptr @.red_comb..2, ptr [[TMP46]], align 8
// CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 6 // CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 6
// CHECK1-NEXT: store i32 1, ptr [[TMP50]], align 8 // CHECK1-NEXT: store i32 1, ptr [[TMP47]], align 8
// CHECK1-NEXT: [[TMP51:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP48:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP52:%.*]] = load i32, ptr [[TMP51]], align 4 // CHECK1-NEXT: [[TMP49:%.*]] = load i32, ptr [[TMP48]], align 4
// CHECK1-NEXT: [[TMP54:%.*]] = call ptr @__kmpc_taskred_modifier_init(ptr @[[GLOB1]], i32 [[TMP52]], i32 0, i32 2, ptr [[DOTRD_INPUT_]]) // CHECK1-NEXT: [[TMP50:%.*]] = call ptr @__kmpc_taskred_modifier_init(ptr @[[GLOB1]], i32 [[TMP49]], i32 0, i32 2, ptr [[DOTRD_INPUT_]])
// CHECK1-NEXT: store ptr [[TMP54]], ptr [[DOTTASK_RED_]], align 8 // CHECK1-NEXT: store ptr [[TMP50]], ptr [[DOTTASK_RED_]], align 8
// CHECK1-NEXT: [[TMP55:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 0 // CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[DOTTASK_RED_]], ptr [[TMP55]], align 8 // CHECK1-NEXT: store ptr [[DOTTASK_RED_]], ptr [[TMP51]], align 8
// CHECK1-NEXT: [[TMP56:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 1 // CHECK1-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP56]], align 8 // CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP52]], align 8
// CHECK1-NEXT: [[TMP57:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 2 // CHECK1-NEXT: [[TMP53:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 2
// CHECK1-NEXT: [[TMP58:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK1-NEXT: [[TMP54:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK1-NEXT: store ptr [[TMP58]], ptr [[TMP57]], align 8 // CHECK1-NEXT: store ptr [[TMP54]], ptr [[TMP53]], align 8
// CHECK1-NEXT: [[TMP59:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP55:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP60:%.*]] = load i32, ptr [[TMP59]], align 4 // CHECK1-NEXT: [[TMP56:%.*]] = load i32, ptr [[TMP55]], align 4
// CHECK1-NEXT: [[TMP61:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP60]], i32 1, i64 48, i64 24, ptr @.omp_task_entry.) // CHECK1-NEXT: [[TMP57:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP56]], i32 1, i64 48, i64 24, ptr @.omp_task_entry.)
// CHECK1-NEXT: [[TMP63:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP61]], i32 0, i32 0 // CHECK1-NEXT: [[TMP58:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP57]], i32 0, i32 0
// CHECK1-NEXT: [[TMP64:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP63]], i32 0, i32 0 // CHECK1-NEXT: [[TMP59:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP58]], i32 0, i32 0
// CHECK1-NEXT: [[TMP65:%.*]] = load ptr, ptr [[TMP64]], align 8 // CHECK1-NEXT: [[TMP60:%.*]] = load ptr, ptr [[TMP59]], align 8
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP65]], ptr align 8 [[AGG_CAPTURED]], i64 24, i1 false) // CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP60]], ptr align 8 [[AGG_CAPTURED]], i64 24, i1 false)
// CHECK1-NEXT: [[TMP67:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP61]], i32 0, i32 1 // CHECK1-NEXT: [[TMP61:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP57]], i32 0, i32 1
// CHECK1-NEXT: [[TMP68:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP67]], i32 0, i32 0 // CHECK1-NEXT: [[TMP62:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP61]], i32 0, i32 0
// CHECK1-NEXT: [[TMP69:%.*]] = load ptr, ptr [[DOTTASK_RED_]], align 8 // CHECK1-NEXT: [[TMP63:%.*]] = load ptr, ptr [[DOTTASK_RED_]], align 8
// CHECK1-NEXT: store ptr [[TMP69]], ptr [[TMP68]], align 8 // CHECK1-NEXT: store ptr [[TMP63]], ptr [[TMP62]], align 8
// CHECK1-NEXT: [[TMP70:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP64:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP71:%.*]] = load i32, ptr [[TMP70]], align 4 // CHECK1-NEXT: [[TMP65:%.*]] = load i32, ptr [[TMP64]], align 4
// CHECK1-NEXT: [[TMP72:%.*]] = call i32 @__kmpc_omp_task(ptr @[[GLOB1]], i32 [[TMP71]], ptr [[TMP61]]) // CHECK1-NEXT: [[TMP66:%.*]] = call i32 @__kmpc_omp_task(ptr @[[GLOB1]], i32 [[TMP65]], ptr [[TMP57]])
// CHECK1-NEXT: [[TMP67:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP68:%.*]] = load i32, ptr [[TMP67]], align 4
// CHECK1-NEXT: call void @__kmpc_task_reduction_modifier_fini(ptr @[[GLOB1]], i32 [[TMP68]], i32 0)
// CHECK1-NEXT: [[TMP69:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP69]], align 8
// CHECK1-NEXT: [[TMP70:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT: store ptr [[VLA]], ptr [[TMP70]], align 8
// CHECK1-NEXT: [[TMP71:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK1-NEXT: [[TMP72:%.*]] = inttoptr i64 [[TMP11]] to ptr
// CHECK1-NEXT: store ptr [[TMP72]], ptr [[TMP71]], align 8
// CHECK1-NEXT: [[TMP73:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP73:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP74:%.*]] = load i32, ptr [[TMP73]], align 4 // CHECK1-NEXT: [[TMP74:%.*]] = load i32, ptr [[TMP73]], align 4
// CHECK1-NEXT: call void @__kmpc_task_reduction_modifier_fini(ptr @[[GLOB1]], i32 [[TMP74]], i32 0) // CHECK1-NEXT: [[TMP75:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB2:[0-9]+]], i32 [[TMP74]], i32 2, i64 24, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: [[TMP75:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: switch i32 [[TMP75]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP75]], align 8
// CHECK1-NEXT: [[TMP77:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT: store ptr [[VLA]], ptr [[TMP77]], align 8
// CHECK1-NEXT: [[TMP78:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK1-NEXT: [[TMP79:%.*]] = inttoptr i64 [[TMP11]] to ptr
// CHECK1-NEXT: store ptr [[TMP79]], ptr [[TMP78]], align 8
// CHECK1-NEXT: [[TMP80:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP81:%.*]] = load i32, ptr [[TMP80]], align 4
// CHECK1-NEXT: [[TMP83:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB2:[0-9]+]], i32 [[TMP81]], i32 2, i64 24, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP83]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP84:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP76:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP85:%.*]] = load i32, ptr [[ARGC1]], align 4 // CHECK1-NEXT: [[TMP77:%.*]] = load i32, ptr [[ARGC1]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP84]], [[TMP85]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP76]], [[TMP77]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP86:%.*]] = getelementptr i8, ptr [[ARRAYIDX2]], i64 [[TMP11]] // CHECK1-NEXT: [[TMP78:%.*]] = getelementptr i8, ptr [[ARRAYIDX2]], i64 [[TMP11]]
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX2]], [[TMP86]] // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX2]], [[TMP78]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE18:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE18:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1: omp.arraycpy.body: // CHECK1: omp.arraycpy.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST12:%.*]] = phi ptr [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT16:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST12:%.*]] = phi ptr [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT16:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[TMP87:%.*]] = load i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST12]], align 1 // CHECK1-NEXT: [[TMP79:%.*]] = load i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST12]], align 1
// CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP87]] to i32 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP79]] to i32
// CHECK1-NEXT: [[TMP88:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 // CHECK1-NEXT: [[TMP80:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
// CHECK1-NEXT: [[CONV13:%.*]] = sext i8 [[TMP88]] to i32 // CHECK1-NEXT: [[CONV13:%.*]] = sext i8 [[TMP80]] to i32
// CHECK1-NEXT: [[ADD14:%.*]] = add nsw i32 [[CONV]], [[CONV13]] // CHECK1-NEXT: [[ADD14:%.*]] = add nsw i32 [[CONV]], [[CONV13]]
// CHECK1-NEXT: [[CONV15:%.*]] = trunc i32 [[ADD14]] to i8 // CHECK1-NEXT: [[CONV15:%.*]] = trunc i32 [[ADD14]] to i8
// CHECK1-NEXT: store i8 [[CONV15]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST12]], align 1 // CHECK1-NEXT: store i8 [[CONV15]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST12]], align 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT16]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST12]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT16]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST12]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE17:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT16]], [[TMP86]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE17:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT16]], [[TMP78]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE17]], label [[OMP_ARRAYCPY_DONE18]], label [[OMP_ARRAYCPY_BODY]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE17]], label [[OMP_ARRAYCPY_DONE18]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1: omp.arraycpy.done18: // CHECK1: omp.arraycpy.done18:
// CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB2]], i32 [[TMP81]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB2]], i32 [[TMP74]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP89:%.*]] = load i32, ptr [[ARGC1]], align 4 // CHECK1-NEXT: [[TMP81:%.*]] = load i32, ptr [[ARGC1]], align 4
// CHECK1-NEXT: [[TMP90:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP89]] monotonic, align 4 // CHECK1-NEXT: [[TMP82:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP81]] monotonic, align 4
// CHECK1-NEXT: [[TMP91:%.*]] = getelementptr i8, ptr [[ARRAYIDX2]], i64 [[TMP11]] // CHECK1-NEXT: [[TMP83:%.*]] = getelementptr i8, ptr [[ARRAYIDX2]], i64 [[TMP11]]
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY19:%.*]] = icmp eq ptr [[ARRAYIDX2]], [[TMP91]] // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY19:%.*]] = icmp eq ptr [[ARRAYIDX2]], [[TMP83]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY19]], label [[OMP_ARRAYCPY_DONE32:%.*]], label [[OMP_ARRAYCPY_BODY20:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY19]], label [[OMP_ARRAYCPY_DONE32:%.*]], label [[OMP_ARRAYCPY_BODY20:%.*]]
// CHECK1: omp.arraycpy.body20: // CHECK1: omp.arraycpy.body20:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST21:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT30:%.*]], [[ATOMIC_EXIT:%.*]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST21:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT30:%.*]], [[ATOMIC_EXIT:%.*]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST22:%.*]] = phi ptr [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT29:%.*]], [[ATOMIC_EXIT]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST22:%.*]] = phi ptr [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT29:%.*]], [[ATOMIC_EXIT]] ]
// CHECK1-NEXT: [[TMP92:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST21]], align 1 // CHECK1-NEXT: [[TMP84:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST21]], align 1
// CHECK1-NEXT: [[CONV23:%.*]] = sext i8 [[TMP92]] to i32 // CHECK1-NEXT: [[CONV23:%.*]] = sext i8 [[TMP84]] to i32
// CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST22]] monotonic, align 1 // CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST22]] monotonic, align 1
// CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]] // CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]]
// CHECK1: atomic_cont: // CHECK1: atomic_cont:
// CHECK1-NEXT: [[TMP93:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY20]] ], [ [[TMP98:%.*]], [[ATOMIC_CONT]] ] // CHECK1-NEXT: [[TMP85:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY20]] ], [ [[TMP90:%.*]], [[ATOMIC_CONT]] ]
// CHECK1-NEXT: store i8 [[TMP93]], ptr [[_TMP24]], align 1 // CHECK1-NEXT: store i8 [[TMP85]], ptr [[_TMP24]], align 1
// CHECK1-NEXT: [[TMP94:%.*]] = load i8, ptr [[_TMP24]], align 1 // CHECK1-NEXT: [[TMP86:%.*]] = load i8, ptr [[_TMP24]], align 1
// CHECK1-NEXT: [[CONV25:%.*]] = sext i8 [[TMP94]] to i32 // CHECK1-NEXT: [[CONV25:%.*]] = sext i8 [[TMP86]] to i32
// CHECK1-NEXT: [[TMP95:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST21]], align 1 // CHECK1-NEXT: [[TMP87:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST21]], align 1
// CHECK1-NEXT: [[CONV26:%.*]] = sext i8 [[TMP95]] to i32 // CHECK1-NEXT: [[CONV26:%.*]] = sext i8 [[TMP87]] to i32
// CHECK1-NEXT: [[ADD27:%.*]] = add nsw i32 [[CONV25]], [[CONV26]] // CHECK1-NEXT: [[ADD27:%.*]] = add nsw i32 [[CONV25]], [[CONV26]]
// CHECK1-NEXT: [[CONV28:%.*]] = trunc i32 [[ADD27]] to i8 // CHECK1-NEXT: [[CONV28:%.*]] = trunc i32 [[ADD27]] to i8
// CHECK1-NEXT: store i8 [[CONV28]], ptr [[ATOMIC_TEMP]], align 1 // CHECK1-NEXT: store i8 [[CONV28]], ptr [[ATOMIC_TEMP]], align 1
// CHECK1-NEXT: [[TMP96:%.*]] = load i8, ptr [[ATOMIC_TEMP]], align 1 // CHECK1-NEXT: [[TMP88:%.*]] = load i8, ptr [[ATOMIC_TEMP]], align 1
// CHECK1-NEXT: [[TMP97:%.*]] = cmpxchg ptr [[OMP_ARRAYCPY_DESTELEMENTPAST22]], i8 [[TMP93]], i8 [[TMP96]] monotonic monotonic, align 1 // CHECK1-NEXT: [[TMP89:%.*]] = cmpxchg ptr [[OMP_ARRAYCPY_DESTELEMENTPAST22]], i8 [[TMP85]], i8 [[TMP88]] monotonic monotonic, align 1
// CHECK1-NEXT: [[TMP98]] = extractvalue { i8, i1 } [[TMP97]], 0 // CHECK1-NEXT: [[TMP90]] = extractvalue { i8, i1 } [[TMP89]], 0
// CHECK1-NEXT: [[TMP99:%.*]] = extractvalue { i8, i1 } [[TMP97]], 1 // CHECK1-NEXT: [[TMP91:%.*]] = extractvalue { i8, i1 } [[TMP89]], 1
// CHECK1-NEXT: br i1 [[TMP99]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]] // CHECK1-NEXT: br i1 [[TMP91]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]]
// CHECK1: atomic_exit: // CHECK1: atomic_exit:
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT29]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST22]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT29]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST22]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT30]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST21]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT30]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST21]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE31:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT29]], [[TMP91]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE31:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT29]], [[TMP83]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE31]], label [[OMP_ARRAYCPY_DONE32]], label [[OMP_ARRAYCPY_BODY20]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE31]], label [[OMP_ARRAYCPY_DONE32]], label [[OMP_ARRAYCPY_BODY20]]
// CHECK1: omp.arraycpy.done32: // CHECK1: omp.arraycpy.done32:
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: [[TMP100:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 // CHECK1-NEXT: [[TMP92:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK1-NEXT: call void @llvm.stackrestore(ptr [[TMP100]]) // CHECK1-NEXT: call void @llvm.stackrestore(ptr [[TMP92]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.red_init. // CHECK1-LABEL: define {{[^@]+}}@.red_init.
// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store i32 0, ptr [[TMP3]], align 4 // CHECK1-NEXT: store i32 0, ptr [[TMP2]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.red_comb. // CHECK1-LABEL: define {{[^@]+}}@.red_comb.
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP3]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP5]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP6]], [[TMP7]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP4]], [[TMP5]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP3]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP2]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.red_init..1 // CHECK1-LABEL: define {{[^@]+}}@.red_init..1
// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
▲ Show 20 Lines • Show All 78 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP3]], i32 0, i32 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP3]], i32 0, i32 0
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP3]], i32 0, i32 1 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP3]], i32 0, i32 1
// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META3:![0-9]+]]) // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META3:![0-9]+]])
// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META6:![0-9]+]]) // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META6:![0-9]+]])
// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META8:![0-9]+]]) // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META8:![0-9]+]])
// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]]) // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]])
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: store ptr [[TMP9]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: store ptr @.omp_task_privates_map., ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr @.omp_task_privates_map., ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: call void [[TMP13]](ptr [[TMP14]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR6]] // CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR6]]
// CHECK1-NEXT: [[TMP16:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], ptr [[TMP12]], i32 0, i32 1 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], ptr [[TMP9]], i32 0, i32 1
// CHECK1-NEXT: [[TMP18:%.*]] = load ptr, ptr [[TMP17]], align 8 // CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = load ptr, ptr [[TMP16]], align 8 // CHECK1-NEXT: [[TMP15:%.*]] = load ptr, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12
// CHECK1-NEXT: [[TMP22:%.*]] = call ptr @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], ptr [[TMP19]], ptr [[TMP18]]) // CHECK1-NEXT: [[TMP17:%.*]] = call ptr @__kmpc_task_reduction_get_th_data(i32 [[TMP16]], ptr [[TMP15]], ptr [[TMP14]])
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP12]], i32 0, i32 2 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP9]], i32 0, i32 2
// CHECK1-NEXT: [[TMP24:%.*]] = load ptr, ptr [[TMP23]], align 8 // CHECK1-NEXT: [[TMP19:%.*]] = load ptr, ptr [[TMP18]], align 8
// CHECK1-NEXT: [[TMP25:%.*]] = load ptr, ptr [[TMP24]], align 8 // CHECK1-NEXT: [[TMP20:%.*]] = load ptr, ptr [[TMP19]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP12]], i32 0, i32 1 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP9]], i32 0, i32 1
// CHECK1-NEXT: [[TMP27:%.*]] = load ptr, ptr [[TMP26]], align 8 // CHECK1-NEXT: [[TMP22:%.*]] = load ptr, ptr [[TMP21]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4 // CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
// CHECK1-NEXT: [[TMP29:%.*]] = sext i32 [[TMP28]] to i64 // CHECK1-NEXT: [[TMP24:%.*]] = sext i32 [[TMP23]] to i64
// CHECK1-NEXT: [[LB_ADD_LEN_I:%.*]] = add nsw i64 -1, [[TMP29]] // CHECK1-NEXT: [[LB_ADD_LEN_I:%.*]] = add nsw i64 -1, [[TMP24]]
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP12]], i32 0, i32 2 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP9]], i32 0, i32 2
// CHECK1-NEXT: [[TMP31:%.*]] = load ptr, ptr [[TMP30]], align 8 // CHECK1-NEXT: [[TMP26:%.*]] = load ptr, ptr [[TMP25]], align 8
// CHECK1-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds ptr, ptr [[TMP31]], i64 9 // CHECK1-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds ptr, ptr [[TMP26]], i64 9
// CHECK1-NEXT: [[TMP32:%.*]] = load ptr, ptr [[ARRAYIDX2_I]], align 8 // CHECK1-NEXT: [[TMP27:%.*]] = load ptr, ptr [[ARRAYIDX2_I]], align 8
// CHECK1-NEXT: [[ARRAYIDX3_I:%.*]] = getelementptr inbounds i8, ptr [[TMP32]], i64 [[LB_ADD_LEN_I]] // CHECK1-NEXT: [[ARRAYIDX3_I:%.*]] = getelementptr inbounds i8, ptr [[TMP27]], i64 [[LB_ADD_LEN_I]]
// CHECK1-NEXT: [[TMP33:%.*]] = ptrtoint ptr [[ARRAYIDX3_I]] to i64 // CHECK1-NEXT: [[TMP28:%.*]] = ptrtoint ptr [[ARRAYIDX3_I]] to i64
// CHECK1-NEXT: [[TMP34:%.*]] = ptrtoint ptr [[TMP25]] to i64 // CHECK1-NEXT: [[TMP29:%.*]] = ptrtoint ptr [[TMP20]] to i64
// CHECK1-NEXT: [[TMP35:%.*]] = sub i64 [[TMP33]], [[TMP34]] // CHECK1-NEXT: [[TMP30:%.*]] = sub i64 [[TMP28]], [[TMP29]]
// CHECK1-NEXT: [[TMP36:%.*]] = sdiv exact i64 [[TMP35]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP31:%.*]] = sdiv exact i64 [[TMP30]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP37:%.*]] = add nuw i64 [[TMP36]], 1 // CHECK1-NEXT: [[TMP32:%.*]] = add nuw i64 [[TMP31]], 1
// CHECK1-NEXT: [[TMP38:%.*]] = mul nuw i64 [[TMP37]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP33:%.*]] = mul nuw i64 [[TMP32]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: store i64 [[TMP37]], ptr @{{reduction_size[.].+[.]}}, align 8, !noalias !12 // CHECK1-NEXT: store i64 [[TMP32]], ptr @{{reduction_size[.].+[.]}}, align 8, !noalias !12
// CHECK1-NEXT: [[TMP39:%.*]] = load ptr, ptr [[TMP16]], align 8 // CHECK1-NEXT: [[TMP34:%.*]] = load ptr, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = call ptr @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], ptr [[TMP39]], ptr [[TMP25]]) // CHECK1-NEXT: [[TMP35:%.*]] = call ptr @__kmpc_task_reduction_get_th_data(i32 [[TMP16]], ptr [[TMP34]], ptr [[TMP20]])
// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP12]], i32 0, i32 2 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP9]], i32 0, i32 2
// CHECK1-NEXT: [[TMP42:%.*]] = load ptr, ptr [[TMP41]], align 8 // CHECK1-NEXT: [[TMP37:%.*]] = load ptr, ptr [[TMP36]], align 8
// CHECK1-NEXT: [[TMP43:%.*]] = load ptr, ptr [[TMP42]], align 8 // CHECK1-NEXT: [[TMP38:%.*]] = load ptr, ptr [[TMP37]], align 8
// CHECK1-NEXT: [[TMP44:%.*]] = ptrtoint ptr [[TMP43]] to i64 // CHECK1-NEXT: [[TMP39:%.*]] = ptrtoint ptr [[TMP38]] to i64
// CHECK1-NEXT: [[TMP45:%.*]] = ptrtoint ptr [[TMP25]] to i64 // CHECK1-NEXT: [[TMP40:%.*]] = ptrtoint ptr [[TMP20]] to i64
// CHECK1-NEXT: [[TMP46:%.*]] = sub i64 [[TMP44]], [[TMP45]] // CHECK1-NEXT: [[TMP41:%.*]] = sub i64 [[TMP39]], [[TMP40]]
// CHECK1-NEXT: [[TMP47:%.*]] = sdiv exact i64 [[TMP46]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP42:%.*]] = sdiv exact i64 [[TMP41]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP48:%.*]] = getelementptr i8, ptr [[TMP40]], i64 [[TMP47]] // CHECK1-NEXT: [[TMP43:%.*]] = getelementptr i8, ptr [[TMP35]], i64 [[TMP42]]
// CHECK1-NEXT: store ptr [[TMP4_I]], ptr [[TMP_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP4_I]], ptr [[TMP_I]], align 8, !noalias !12
// CHECK1-NEXT: store ptr [[TMP48]], ptr [[TMP4_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP43]], ptr [[TMP4_I]], align 8, !noalias !12
// CHECK1-NEXT: ret i32 0 // CHECK1-NEXT: ret i32 0
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP3]], i64 0, i64 1
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP4]], i64 0, i64 1 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 1
// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 2
// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[TMP12]], align 8 // CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 1 // CHECK1-NEXT: [[TMP14:%.*]] = ptrtoint ptr [[TMP13]] to i64
// CHECK1-NEXT: [[TMP15:%.*]] = load ptr, ptr [[TMP14]], align 8 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 2 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP17:%.*]] = load ptr, ptr [[TMP16]], align 8 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP15]], [[TMP16]]
// CHECK1-NEXT: [[TMP18:%.*]] = ptrtoint ptr [[TMP17]] to i64 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[TMP11]], i64 [[TMP14]]
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP11]], [[TMP17]]
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[TMP15]], i64 [[TMP18]]
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP15]], [[TMP21]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1: omp.arraycpy.body: // CHECK1: omp.arraycpy.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP13]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP9]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP15]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[TMP22:%.*]] = load i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 // CHECK1-NEXT: [[TMP18:%.*]] = load i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP22]] to i32 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP18]] to i32
// CHECK1-NEXT: [[TMP23:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 // CHECK1-NEXT: [[TMP19:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
// CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP23]] to i32 // CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP19]] to i32
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]] // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]]
// CHECK1-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8 // CHECK1-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8
// CHECK1-NEXT: store i8 [[CONV4]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 // CHECK1-NEXT: store i8 [[CONV4]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP21]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP17]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1: omp.arraycpy.done5: // CHECK1: omp.arraycpy.done5:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //

clang/test/OpenMP/target_teams_distribute_collapse_codegen.cpp

Show First 20 Lines • Show All 110 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 8 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[A]], ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr [[A]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 56088, ptr [[TMP15]], align 8 // CHECK1-NEXT: store i64 56088, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP15]], 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]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 // CHECK1-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i64 0, i64 0 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i64 0, i64 0
// CHECK1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 0 // CHECK1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 0
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4
// CHECK1-NEXT: ret i32 [[TMP18]] // CHECK1-NEXT: ret i32 [[TMP16]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28
// CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
▲ Show 20 Lines • Show All 107 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[A]], ptr [[TMP2]], align 4 // CHECK3-NEXT: store ptr [[A]], ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 56088, ptr [[TMP15]], align 8 // CHECK3-NEXT: store i64 56088, ptr [[TMP13]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP15]], 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]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 // CHECK3-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i32 0, i32 0 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i32 0, i32 0
// CHECK3-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 0 // CHECK3-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 0
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4
// CHECK3-NEXT: ret i32 [[TMP18]] // CHECK3-NEXT: ret i32 [[TMP16]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28
// CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
▲ Show 20 Lines • Show All 101 Lines • ▼ Show 20 Lines
// 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 ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], 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: [[_TMP1:%.*]] = 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_2:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK9-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8
// CHECK9-NEXT: store i32 100, ptr [[N]], align 4 // CHECK9-NEXT: store i32 100, ptr [[N]], align 4
// CHECK9-NEXT: store i32 2, ptr [[M]], align 4 // CHECK9-NEXT: store i32 2, ptr [[M]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[M]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64 // CHECK9-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
// CHECK9-NEXT: [[TMP4:%.*]] = call ptr @llvm.stacksave() // CHECK9-NEXT: [[TMP4:%.*]] = call ptr @llvm.stacksave()
// CHECK9-NEXT: store ptr [[TMP4]], ptr [[SAVED_STACK]], align 8 // CHECK9-NEXT: store ptr [[TMP4]], ptr [[SAVED_STACK]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP5]], align 4 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP5]], align 4
// CHECK9-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8
// CHECK9-NEXT: store i64 [[TMP3]], ptr [[__VLA_EXPR1]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], ptr [[__VLA_EXPR1]], align 8
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[N]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP6]], ptr [[N_CASTED]], align 4 // CHECK9-NEXT: store i32 [[TMP6]], ptr [[N_CASTED]], align 4
// CHECK9-NEXT: [[TMP7:%.*]] = load i64, ptr [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP7:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[M]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP8]], ptr [[M_CASTED]], align 4 // CHECK9-NEXT: store i32 [[TMP8]], ptr [[M_CASTED]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[M_CASTED]], align 8 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[M_CASTED]], align 8
// 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: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 40, i1 false) // CHECK9-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 40, i1 false)
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP13]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP13]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP15]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP14]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP16]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store ptr null, ptr [[TMP17]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP17]], align 8
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP18]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP18]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP20]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP19]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: store ptr null, ptr [[TMP22]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP20]], align 8
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP23]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP21]], align 8
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP25]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP22]], align 8
// CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK9-NEXT: store ptr null, ptr [[TMP23]], align 8
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP24]], align 8
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP25]], align 8
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK9-NEXT: store i64 [[TMP11]], ptr [[TMP26]], align 8
// CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK9-NEXT: store ptr null, ptr [[TMP27]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP27]], align 8
// CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP28]], align 8 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP30]], align 8 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK9-NEXT: store i32 [[TMP31]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: store ptr null, ptr [[TMP32]], align 8 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK9-NEXT: store i32 [[TMP32]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP33]], align 8 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP33]], 0
// CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP35]], align 8
// CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK9-NEXT: store i64 [[TMP11]], ptr [[TMP37]], align 8
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK9-NEXT: store ptr null, ptr [[TMP38]], align 8
// CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP42:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP42]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP43:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP43]], ptr [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP44:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP44]], 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: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP45:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4 // CHECK9-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP45]], 0 // CHECK9-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP34]], 0
// CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK9-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK9-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB9]], ptr [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: store i64 [[SUB7]], ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK9-NEXT: [[TMP46:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: [[TMP35:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP46]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP35]], 1
// CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK9-NEXT: store i32 1, ptr [[TMP47]], align 4 // CHECK9-NEXT: store i32 1, ptr [[TMP36]], align 4
// CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK9-NEXT: store i32 5, ptr [[TMP48]], align 4 // CHECK9-NEXT: store i32 5, ptr [[TMP37]], align 4
// CHECK9-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK9-NEXT: store ptr [[TMP39]], ptr [[TMP49]], align 8 // CHECK9-NEXT: store ptr [[TMP28]], ptr [[TMP38]], align 8
// CHECK9-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK9-NEXT: store ptr [[TMP40]], ptr [[TMP50]], align 8 // CHECK9-NEXT: store ptr [[TMP29]], ptr [[TMP39]], align 8
// CHECK9-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK9-NEXT: store ptr [[TMP41]], ptr [[TMP51]], align 8 // CHECK9-NEXT: store ptr [[TMP30]], ptr [[TMP40]], align 8
// CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK9-NEXT: store ptr @.offload_maptypes, ptr [[TMP52]], align 8 // CHECK9-NEXT: store ptr @.offload_maptypes, ptr [[TMP41]], align 8
// CHECK9-NEXT: [[TMP53:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK9-NEXT: store ptr null, ptr [[TMP53]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP42]], align 8
// CHECK9-NEXT: [[TMP54:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK9-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK9-NEXT: store ptr null, ptr [[TMP54]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP43]], align 8
// CHECK9-NEXT: [[TMP55:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK9-NEXT: store i64 [[ADD]], ptr [[TMP55]], align 8 // CHECK9-NEXT: store i64 [[ADD]], ptr [[TMP44]], align 8
// CHECK9-NEXT: [[TMP56:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, ptr [[KERNEL_ARGS]]) // CHECK9-NEXT: [[TMP45:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, ptr [[KERNEL_ARGS]])
// CHECK9-NEXT: [[TMP57:%.*]] = icmp ne i32 [[TMP56]], 0 // CHECK9-NEXT: [[TMP46:%.*]] = icmp ne i32 [[TMP45]], 0
// CHECK9-NEXT: br i1 [[TMP57]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP46]], 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]], ptr [[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]], ptr [[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: [[TMP58:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK9-NEXT: [[TMP47:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef signext [[TMP58]]) // CHECK9-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef signext [[TMP47]])
// CHECK9-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP59:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP48:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(ptr [[TMP59]]) // CHECK9-NEXT: call void @llvm.stackrestore(ptr [[TMP48]])
// CHECK9-NEXT: [[TMP60:%.*]] = load i32, ptr [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP49:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP60]] // CHECK9-NEXT: ret i32 [[TMP49]]
// //
// //
// 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 noundef [[N:%.*]], i64 noundef [[M:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[M:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef 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 All 26 Lines
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// 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
// CHECK9-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP3:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_5:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_6:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_5:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[J:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[J:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I13:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I11:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[J14:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[J12:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8 // CHECK9-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8
// CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT: store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8 // CHECK9-NEXT: store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[CONV7:%.*]] = sext i32 [[DIV]] to i64 // CHECK9-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP6]], 0 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP6]], 0
// CHECK9-NEXT: [[DIV9:%.*]] = sdiv i32 [[SUB8]], 1 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK9-NEXT: [[CONV10:%.*]] = sext i32 [[DIV9]] to i64 // CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV7]], [[CONV10]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV8]]
// CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB11]], ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: store i64 [[SUB9]], ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: store i32 0, ptr [[I]], align 4 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4
// CHECK9-NEXT: store i32 0, ptr [[J]], align 4 // CHECK9-NEXT: store i32 0, ptr [[J]], align 4
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]] // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]] // CHECK9-NEXT: br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK9: land.lhs.true: // CHECK9: land.lhs.true:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[CMP12:%.*]] = icmp slt i32 0, [[TMP8]] // CHECK9-NEXT: [[CMP10:%.*]] = icmp slt i32 0, [[TMP8]]
// CHECK9-NEXT: br i1 [[CMP12]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]] // CHECK9-NEXT: br i1 [[CMP10]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]]
// CHECK9: omp.precond.then: // CHECK9: omp.precond.then:
// CHECK9-NEXT: store i64 0, ptr [[DOTOMP_LB]], align 8 // CHECK9-NEXT: store i64 0, ptr [[DOTOMP_LB]], align 8
// CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: store i64 [[TMP9]], ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: store i64 1, ptr [[DOTOMP_STRIDE]], align 8 // CHECK9-NEXT: store i64 1, ptr [[DOTOMP_STRIDE]], align 8
// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB1:[0-9]+]], i32 [[TMP11]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1) // CHECK9-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB1:[0-9]+]], i32 [[TMP11]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK9-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: [[CMP15:%.*]] = icmp sgt i64 [[TMP12]], [[TMP13]] // CHECK9-NEXT: [[CMP13:%.*]] = icmp sgt i64 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: br i1 [[CMP15]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9-NEXT: br i1 [[CMP13]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true: // CHECK9: cond.true:
// CHECK9-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false: // CHECK9: cond.false:
// CHECK9-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: br label [[COND_END]] // CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end: // CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i64 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] // CHECK9-NEXT: [[COND:%.*]] = phi i64 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i64 [[COND]], ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: store i64 [[COND]], ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8 // CHECK9-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK9-NEXT: store i64 [[TMP16]], ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: store i64 [[TMP16]], ptr [[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: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: [[CMP16:%.*]] = icmp sle i64 [[TMP17]], [[TMP18]] // CHECK9-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP17]], [[TMP18]]
// CHECK9-NEXT: br i1 [[CMP16]], 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: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[SUB17:%.*]] = sub nsw i32 [[TMP20]], 0 // CHECK9-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP20]], 0
// CHECK9-NEXT: [[DIV18:%.*]] = sdiv i32 [[SUB17]], 1 // CHECK9-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1
// CHECK9-NEXT: [[MUL19:%.*]] = mul nsw i32 1, [[DIV18]] // CHECK9-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]]
// CHECK9-NEXT: [[CONV20:%.*]] = sext i32 [[MUL19]] to i64 // CHECK9-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64
// CHECK9-NEXT: [[DIV21:%.*]] = sdiv i64 [[TMP19]], [[CONV20]] // CHECK9-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP19]], [[CONV18]]
// CHECK9-NEXT: [[MUL22:%.*]] = mul nsw i64 [[DIV21]], 1 // CHECK9-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL22]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]]
// CHECK9-NEXT: [[CONV23:%.*]] = trunc i64 [[ADD]] to i32 // CHECK9-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32
// CHECK9-NEXT: store i32 [[CONV23]], ptr [[I13]], align 4 // CHECK9-NEXT: store i32 [[CONV21]], ptr [[I11]], align 4
// CHECK9-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP22:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP23]], 0 // CHECK9-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP23]], 0
// CHECK9-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1 // CHECK9-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1
// CHECK9-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]] // CHECK9-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]]
// CHECK9-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64 // CHECK9-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64
// CHECK9-NEXT: [[DIV28:%.*]] = sdiv i64 [[TMP22]], [[CONV27]] // CHECK9-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP22]], [[CONV25]]
// CHECK9-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP24]], 0 // CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK9-NEXT: [[DIV30:%.*]] = sdiv i32 [[SUB29]], 1 // CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK9-NEXT: [[MUL31:%.*]] = mul nsw i32 1, [[DIV30]] // CHECK9-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]]
// CHECK9-NEXT: [[CONV32:%.*]] = sext i32 [[MUL31]] to i64 // CHECK9-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64
// CHECK9-NEXT: [[MUL33:%.*]] = mul nsw i64 [[DIV28]], [[CONV32]] // CHECK9-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]]
// CHECK9-NEXT: [[SUB34:%.*]] = sub nsw i64 [[TMP21]], [[MUL33]] // CHECK9-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP21]], [[MUL31]]
// CHECK9-NEXT: [[MUL35:%.*]] = mul nsw i64 [[SUB34]], 1 // CHECK9-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1
// CHECK9-NEXT: [[ADD36:%.*]] = add nsw i64 0, [[MUL35]] // CHECK9-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]]
// CHECK9-NEXT: [[CONV37:%.*]] = trunc i64 [[ADD36]] to i32 // CHECK9-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32
// CHECK9-NEXT: store i32 [[CONV37]], ptr [[J14]], align 4 // CHECK9-NEXT: store i32 [[CONV35]], ptr [[J12]], align 4
// CHECK9-NEXT: [[TMP25:%.*]] = load i32, ptr [[I13]], align 4 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, ptr [[I11]], align 4
// CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP25]] to i64 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP25]] to i64
// CHECK9-NEXT: [[TMP26:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP1]] // CHECK9-NEXT: [[TMP26:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP1]]
// CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[TMP26]] // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[TMP26]]
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, ptr [[J14]], align 4 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, ptr [[J12]], align 4
// CHECK9-NEXT: [[IDXPROM38:%.*]] = sext i32 [[TMP27]] to i64 // CHECK9-NEXT: [[IDXPROM36:%.*]] = sext i32 [[TMP27]] to i64
// CHECK9-NEXT: [[ARRAYIDX39:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i64 [[IDXPROM38]] // CHECK9-NEXT: [[ARRAYIDX37:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i64 [[IDXPROM36]]
// CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX39]], align 4 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX37]], align 4
// 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: [[TMP28:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP28:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT: [[ADD40:%.*]] = add nsw i64 [[TMP28]], 1 // CHECK9-NEXT: [[ADD38:%.*]] = add nsw i64 [[TMP28]], 1
// CHECK9-NEXT: store i64 [[ADD40]], ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: store i64 [[ADD38]], ptr [[DOTOMP_IV]], align 8
// 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: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP30]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP30]])
// CHECK9-NEXT: br label [[OMP_PRECOND_END]] // CHECK9-NEXT: br label [[OMP_PRECOND_END]]
Show All 9 Lines
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store ptr [[A]], ptr [[TMP0]], align 8 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP0]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: store ptr [[A]], ptr [[TMP2]], align 8 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP1]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK9-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK9-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK9-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK9-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK9-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8 // CHECK9-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK9-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK9-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK9-NEXT: store ptr @.offload_sizes.2, ptr [[TMP11]], align 8 // CHECK9-NEXT: store ptr @.offload_sizes.2, ptr [[TMP9]], align 8
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK9-NEXT: store ptr @.offload_maptypes.3, ptr [[TMP12]], align 8 // CHECK9-NEXT: store ptr @.offload_maptypes.3, ptr [[TMP10]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK9-NEXT: store ptr null, ptr [[TMP13]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK9-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK9-NEXT: store i64 20, ptr [[TMP15]], align 8 // CHECK9-NEXT: store i64 20, ptr [[TMP13]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, ptr [[KERNEL_ARGS]]) // CHECK9-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, ptr [[KERNEL_ARGS]])
// CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP15]], 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(ptr [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67(ptr [[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
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67
▲ Show 20 Lines • Show All 134 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[M]], align 4 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP6]], ptr [[M_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP6]], ptr [[M_CASTED]], align 4
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[M_CASTED]], align 4 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[M_CASTED]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP0]], [[TMP1]] // CHECK11-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP0]], [[TMP1]]
// 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: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 40, i1 false) // CHECK11-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 40, i1 false)
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP11]], align 4
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP12]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP12]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP14]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP13]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store ptr null, ptr [[TMP16]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP16]], align 4
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP17]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP17]], align 4
// CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP19]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP18]], align 4
// CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: store ptr null, ptr [[TMP21]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP19]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP22]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP20]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP24]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP21]], align 4
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK11-NEXT: store ptr null, ptr [[TMP22]], align 4
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP23]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP24]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK11-NEXT: store i64 [[TMP10]], ptr [[TMP25]], align 4
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK11-NEXT: store ptr null, ptr [[TMP26]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP26]], align 4
// CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP27]], align 4 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP29]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK11-NEXT: store i32 [[TMP30]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store ptr null, ptr [[TMP31]], align 4 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, ptr [[M]], align 4
// CHECK11-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK11-NEXT: store i32 [[TMP31]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP32]], align 4 // CHECK11-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP32]], 0
// CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP34]], align 4
// CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK11-NEXT: store i64 [[TMP10]], ptr [[TMP36]], align 4
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK11-NEXT: store ptr null, ptr [[TMP37]], align 4
// CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP41:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP41]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = load i32, ptr [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP42]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP43:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP43]], 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: [[TMP44:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP44]], 0 // CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP33]], 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]], ptr [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: store i64 [[SUB7]], ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[TMP45:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: [[TMP34:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP45]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP34]], 1
// CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK11-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK11-NEXT: store i32 1, ptr [[TMP46]], align 4 // CHECK11-NEXT: store i32 1, ptr [[TMP35]], align 4
// CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK11-NEXT: store i32 5, ptr [[TMP47]], align 4 // CHECK11-NEXT: store i32 5, ptr [[TMP36]], align 4
// CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK11-NEXT: store ptr [[TMP38]], ptr [[TMP48]], align 4 // CHECK11-NEXT: store ptr [[TMP27]], ptr [[TMP37]], align 4
// CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK11-NEXT: store ptr [[TMP39]], ptr [[TMP49]], align 4 // CHECK11-NEXT: store ptr [[TMP28]], ptr [[TMP38]], align 4
// CHECK11-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK11-NEXT: store ptr [[TMP40]], ptr [[TMP50]], align 4 // CHECK11-NEXT: store ptr [[TMP29]], ptr [[TMP39]], align 4
// CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK11-NEXT: store ptr @.offload_maptypes, ptr [[TMP51]], align 4 // CHECK11-NEXT: store ptr @.offload_maptypes, ptr [[TMP40]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK11-NEXT: store ptr null, ptr [[TMP52]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP41]], align 4
// CHECK11-NEXT: [[TMP53:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK11-NEXT: store ptr null, ptr [[TMP53]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP42]], align 4
// CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK11-NEXT: store i64 [[ADD]], ptr [[TMP54]], align 8 // CHECK11-NEXT: store i64 [[ADD]], ptr [[TMP43]], align 8
// CHECK11-NEXT: [[TMP55:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, ptr [[KERNEL_ARGS]]) // CHECK11-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, ptr [[KERNEL_ARGS]])
// CHECK11-NEXT: [[TMP56:%.*]] = icmp ne i32 [[TMP55]], 0 // CHECK11-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK11-NEXT: br i1 [[TMP56]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP45]], 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]], ptr [[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]], ptr [[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: [[TMP57:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK11-NEXT: [[TMP46:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef [[TMP57]]) // CHECK11-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef [[TMP46]])
// CHECK11-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP58:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(ptr [[TMP58]]) // CHECK11-NEXT: call void @llvm.stackrestore(ptr [[TMP47]])
// CHECK11-NEXT: [[TMP59:%.*]] = load i32, ptr [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP48:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP59]] // CHECK11-NEXT: ret i32 [[TMP48]]
// //
// //
// 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 noundef [[N:%.*]], i32 noundef [[M:%.*]], i32 noundef [[VLA:%.*]], i32 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[M:%.*]], i32 noundef [[VLA:%.*]], i32 noundef [[VLA1:%.*]], ptr noundef 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 171 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], 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: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store ptr [[A]], ptr [[TMP0]], align 4 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP0]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: store ptr [[A]], ptr [[TMP2]], align 4 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP1]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK11-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK11-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK11-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK11-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK11-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 // CHECK11-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK11-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK11-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK11-NEXT: store ptr @.offload_sizes.2, ptr [[TMP11]], align 4 // CHECK11-NEXT: store ptr @.offload_sizes.2, ptr [[TMP9]], align 4
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK11-NEXT: store ptr @.offload_maptypes.3, ptr [[TMP12]], align 4 // CHECK11-NEXT: store ptr @.offload_maptypes.3, ptr [[TMP10]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK11-NEXT: store ptr null, ptr [[TMP13]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK11-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK11-NEXT: store i64 20, ptr [[TMP15]], align 8 // CHECK11-NEXT: store i64 20, ptr [[TMP13]], align 8
// CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, ptr [[KERNEL_ARGS]]) // CHECK11-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, ptr [[KERNEL_ARGS]])
// CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK11-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP15]], 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(ptr [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67(ptr [[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
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67
▲ Show 20 Lines • Show All 94 Lines • Show Last 20 Lines

clang/test/OpenMP/target_teams_distribute_parallel_for_collapse_codegen.cpp

Show First 20 Lines • Show All 115 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 8 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[A]], ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr [[A]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 56088, ptr [[TMP15]], align 8 // CHECK1-NEXT: store i64 56088, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP15]], 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]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 // CHECK1-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i64 0, i64 0 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i64 0, i64 0
// CHECK1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 0 // CHECK1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 0
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4
// CHECK1-NEXT: ret i32 [[TMP18]] // CHECK1-NEXT: ret i32 [[TMP16]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28
// CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
▲ Show 20 Lines • Show All 182 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[A]], ptr [[TMP2]], align 4 // CHECK3-NEXT: store ptr [[A]], ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 56088, ptr [[TMP15]], align 8 // CHECK3-NEXT: store i64 56088, ptr [[TMP13]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP15]], 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]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 // CHECK3-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i32 0, i32 0 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i32 0, i32 0
// CHECK3-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 0 // CHECK3-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 0
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4
// CHECK3-NEXT: ret i32 [[TMP18]] // CHECK3-NEXT: ret i32 [[TMP16]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28
// CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
▲ Show 20 Lines • Show All 172 Lines • ▼ Show 20 Lines
// 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 ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], 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: [[_TMP1:%.*]] = 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_2:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK9-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8
// CHECK9-NEXT: store i32 100, ptr [[N]], align 4 // CHECK9-NEXT: store i32 100, ptr [[N]], align 4
// CHECK9-NEXT: store i32 2, ptr [[M]], align 4 // CHECK9-NEXT: store i32 2, ptr [[M]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[M]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64 // CHECK9-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
// CHECK9-NEXT: [[TMP4:%.*]] = call ptr @llvm.stacksave() // CHECK9-NEXT: [[TMP4:%.*]] = call ptr @llvm.stacksave()
// CHECK9-NEXT: store ptr [[TMP4]], ptr [[SAVED_STACK]], align 8 // CHECK9-NEXT: store ptr [[TMP4]], ptr [[SAVED_STACK]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP5]], align 4 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP5]], align 4
// CHECK9-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8
// CHECK9-NEXT: store i64 [[TMP3]], ptr [[__VLA_EXPR1]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], ptr [[__VLA_EXPR1]], align 8
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[N]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP6]], ptr [[N_CASTED]], align 4 // CHECK9-NEXT: store i32 [[TMP6]], ptr [[N_CASTED]], align 4
// CHECK9-NEXT: [[TMP7:%.*]] = load i64, ptr [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP7:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[M]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP8]], ptr [[M_CASTED]], align 4 // CHECK9-NEXT: store i32 [[TMP8]], ptr [[M_CASTED]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[M_CASTED]], align 8 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[M_CASTED]], align 8
// 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: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 40, i1 false) // CHECK9-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 40, i1 false)
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP13]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP13]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP15]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP14]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP16]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store ptr null, ptr [[TMP17]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP17]], align 8
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP18]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP18]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP20]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP19]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: store ptr null, ptr [[TMP22]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP20]], align 8
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP23]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP21]], align 8
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP25]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP22]], align 8
// CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK9-NEXT: store ptr null, ptr [[TMP23]], align 8
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP24]], align 8
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP25]], align 8
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK9-NEXT: store i64 [[TMP11]], ptr [[TMP26]], align 8
// CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK9-NEXT: store ptr null, ptr [[TMP27]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP27]], align 8
// CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP28]], align 8 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP30]], align 8 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK9-NEXT: store i32 [[TMP31]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: store ptr null, ptr [[TMP32]], align 8 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK9-NEXT: store i32 [[TMP32]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP33]], align 8 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP33]], 0
// CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP35]], align 8
// CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK9-NEXT: store i64 [[TMP11]], ptr [[TMP37]], align 8
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK9-NEXT: store ptr null, ptr [[TMP38]], align 8
// CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP42:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP42]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP43:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP43]], ptr [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP44:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP44]], 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: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP45:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4 // CHECK9-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP45]], 0 // CHECK9-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP34]], 0
// CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK9-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK9-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB9]], ptr [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: store i64 [[SUB7]], ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK9-NEXT: [[TMP46:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: [[TMP35:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP46]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP35]], 1
// CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK9-NEXT: store i32 1, ptr [[TMP47]], align 4 // CHECK9-NEXT: store i32 1, ptr [[TMP36]], align 4
// CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK9-NEXT: store i32 5, ptr [[TMP48]], align 4 // CHECK9-NEXT: store i32 5, ptr [[TMP37]], align 4
// CHECK9-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK9-NEXT: store ptr [[TMP39]], ptr [[TMP49]], align 8 // CHECK9-NEXT: store ptr [[TMP28]], ptr [[TMP38]], align 8
// CHECK9-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK9-NEXT: store ptr [[TMP40]], ptr [[TMP50]], align 8 // CHECK9-NEXT: store ptr [[TMP29]], ptr [[TMP39]], align 8
// CHECK9-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK9-NEXT: store ptr [[TMP41]], ptr [[TMP51]], align 8 // CHECK9-NEXT: store ptr [[TMP30]], ptr [[TMP40]], align 8
// CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK9-NEXT: store ptr @.offload_maptypes, ptr [[TMP52]], align 8 // CHECK9-NEXT: store ptr @.offload_maptypes, ptr [[TMP41]], align 8
// CHECK9-NEXT: [[TMP53:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK9-NEXT: store ptr null, ptr [[TMP53]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP42]], align 8
// CHECK9-NEXT: [[TMP54:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK9-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK9-NEXT: store ptr null, ptr [[TMP54]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP43]], align 8
// CHECK9-NEXT: [[TMP55:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK9-NEXT: store i64 [[ADD]], ptr [[TMP55]], align 8 // CHECK9-NEXT: store i64 [[ADD]], ptr [[TMP44]], align 8
// CHECK9-NEXT: [[TMP56:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, ptr [[KERNEL_ARGS]]) // CHECK9-NEXT: [[TMP45:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, ptr [[KERNEL_ARGS]])
// CHECK9-NEXT: [[TMP57:%.*]] = icmp ne i32 [[TMP56]], 0 // CHECK9-NEXT: [[TMP46:%.*]] = icmp ne i32 [[TMP45]], 0
// CHECK9-NEXT: br i1 [[TMP57]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP46]], 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]], ptr [[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]], ptr [[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: [[TMP58:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK9-NEXT: [[TMP47:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef signext [[TMP58]]) // CHECK9-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef signext [[TMP47]])
// CHECK9-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP59:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP48:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(ptr [[TMP59]]) // CHECK9-NEXT: call void @llvm.stackrestore(ptr [[TMP48]])
// CHECK9-NEXT: [[TMP60:%.*]] = load i32, ptr [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP49:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP60]] // CHECK9-NEXT: ret i32 [[TMP49]]
// //
// //
// 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 noundef [[N:%.*]], i64 noundef [[M:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[M:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef 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 All 26 Lines
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// 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
// CHECK9-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP3:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_5:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_6:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_5:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[J:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[J:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I13:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I11:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[J14:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[J12:%.*]] = alloca i32, align 4
// 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: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8 // CHECK9-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8
// CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT: store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8 // CHECK9-NEXT: store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[CONV7:%.*]] = sext i32 [[DIV]] to i64 // CHECK9-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP6]], 0 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP6]], 0
// CHECK9-NEXT: [[DIV9:%.*]] = sdiv i32 [[SUB8]], 1 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK9-NEXT: [[CONV10:%.*]] = sext i32 [[DIV9]] to i64 // CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV7]], [[CONV10]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV8]]
// CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB11]], ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: store i64 [[SUB9]], ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: store i32 0, ptr [[I]], align 4 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4
// CHECK9-NEXT: store i32 0, ptr [[J]], align 4 // CHECK9-NEXT: store i32 0, ptr [[J]], align 4
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]] // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]] // CHECK9-NEXT: br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK9: land.lhs.true: // CHECK9: land.lhs.true:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[CMP12:%.*]] = icmp slt i32 0, [[TMP8]] // CHECK9-NEXT: [[CMP10:%.*]] = icmp slt i32 0, [[TMP8]]
// CHECK9-NEXT: br i1 [[CMP12]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]] // CHECK9-NEXT: br i1 [[CMP10]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]]
// CHECK9: omp.precond.then: // CHECK9: omp.precond.then:
// CHECK9-NEXT: store i64 0, ptr [[DOTOMP_COMB_LB]], align 8 // CHECK9-NEXT: store i64 0, ptr [[DOTOMP_COMB_LB]], align 8
// CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[DOTOMP_COMB_UB]], align 8 // CHECK9-NEXT: store i64 [[TMP9]], ptr [[DOTOMP_COMB_UB]], align 8
// CHECK9-NEXT: store i64 1, ptr [[DOTOMP_STRIDE]], align 8 // CHECK9-NEXT: store i64 1, ptr [[DOTOMP_STRIDE]], align 8
// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB1:[0-9]+]], i32 [[TMP11]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1) // CHECK9-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB1:[0-9]+]], i32 [[TMP11]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK9-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8 // CHECK9-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: [[CMP15:%.*]] = icmp sgt i64 [[TMP12]], [[TMP13]] // CHECK9-NEXT: [[CMP13:%.*]] = icmp sgt i64 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: br i1 [[CMP15]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9-NEXT: br i1 [[CMP13]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true: // CHECK9: cond.true:
// CHECK9-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false: // CHECK9: cond.false:
// CHECK9-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8 // CHECK9-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK9-NEXT: br label [[COND_END]] // CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end: // CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i64 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] // CHECK9-NEXT: [[COND:%.*]] = phi i64 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i64 [[COND]], ptr [[DOTOMP_COMB_UB]], align 8 // CHECK9-NEXT: store i64 [[COND]], ptr [[DOTOMP_COMB_UB]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8 // CHECK9-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
// CHECK9-NEXT: store i64 [[TMP16]], ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: store i64 [[TMP16]], ptr [[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: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK9-NEXT: [[CMP16:%.*]] = icmp sle i64 [[TMP17]], [[TMP18]] // CHECK9-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP17]], [[TMP18]]
// CHECK9-NEXT: br i1 [[CMP16]], 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: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8 // CHECK9-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8 // CHECK9-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK9-NEXT: store i32 [[TMP21]], ptr [[N_CASTED]], align 4 // CHECK9-NEXT: store i32 [[TMP21]], ptr [[N_CASTED]], align 4
// CHECK9-NEXT: [[TMP22:%.*]] = load i64, ptr [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP22:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP23:%.*]] = load i32, ptr [[M_ADDR]], align 4 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK9-NEXT: store i32 [[TMP23]], ptr [[M_CASTED]], align 4 // CHECK9-NEXT: store i32 [[TMP23]], ptr [[M_CASTED]], align 4
Show All 26 Lines
// 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: [[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
// CHECK9-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP3:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_5:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_6:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_5:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[J:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[J:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I13:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I11:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[J14:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[J12:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8 // CHECK9-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8
// CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT: store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8 // CHECK9-NEXT: store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[CONV7:%.*]] = sext i32 [[DIV]] to i64 // CHECK9-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP6]], 0 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP6]], 0
// CHECK9-NEXT: [[DIV9:%.*]] = sdiv i32 [[SUB8]], 1 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK9-NEXT: [[CONV10:%.*]] = sext i32 [[DIV9]] to i64 // CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV7]], [[CONV10]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV8]]
// CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB11]], ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: store i64 [[SUB9]], ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: store i32 0, ptr [[I]], align 4 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4
// CHECK9-NEXT: store i32 0, ptr [[J]], align 4 // CHECK9-NEXT: store i32 0, ptr [[J]], align 4
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]] // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]] // CHECK9-NEXT: br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK9: land.lhs.true: // CHECK9: land.lhs.true:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[CMP12:%.*]] = icmp slt i32 0, [[TMP8]] // CHECK9-NEXT: [[CMP10:%.*]] = icmp slt i32 0, [[TMP8]]
// CHECK9-NEXT: br i1 [[CMP12]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]] // CHECK9-NEXT: br i1 [[CMP10]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]]
// CHECK9: omp.precond.then: // CHECK9: omp.precond.then:
// CHECK9-NEXT: store i64 0, ptr [[DOTOMP_LB]], align 8 // CHECK9-NEXT: store i64 0, ptr [[DOTOMP_LB]], align 8
// CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: store i64 [[TMP9]], ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: [[TMP10:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: [[TMP11:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT: store i64 [[TMP10]], ptr [[DOTOMP_LB]], align 8 // CHECK9-NEXT: store i64 [[TMP10]], ptr [[DOTOMP_LB]], align 8
// CHECK9-NEXT: store i64 [[TMP11]], ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: store i64 [[TMP11]], ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: store i64 1, ptr [[DOTOMP_STRIDE]], align 8 // CHECK9-NEXT: store i64 1, ptr [[DOTOMP_STRIDE]], align 8
// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB2:[0-9]+]], i32 [[TMP13]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1) // CHECK9-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB2:[0-9]+]], i32 [[TMP13]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK9-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: [[CMP15:%.*]] = icmp sgt i64 [[TMP14]], [[TMP15]] // CHECK9-NEXT: [[CMP13:%.*]] = icmp sgt i64 [[TMP14]], [[TMP15]]
// CHECK9-NEXT: br i1 [[CMP15]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9-NEXT: br i1 [[CMP13]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true: // CHECK9: cond.true:
// CHECK9-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false: // CHECK9: cond.false:
// CHECK9-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// 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]], ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: store i64 [[COND]], ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK9-NEXT: store i64 [[TMP18]], ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: store i64 [[TMP18]], ptr [[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, ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: [[CMP16:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]] // CHECK9-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]]
// CHECK9-NEXT: br i1 [[CMP16]], 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, ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[SUB17:%.*]] = sub nsw i32 [[TMP22]], 0 // CHECK9-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP22]], 0
// CHECK9-NEXT: [[DIV18:%.*]] = sdiv i32 [[SUB17]], 1 // CHECK9-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1
// CHECK9-NEXT: [[MUL19:%.*]] = mul nsw i32 1, [[DIV18]] // CHECK9-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]]
// CHECK9-NEXT: [[CONV20:%.*]] = sext i32 [[MUL19]] to i64 // CHECK9-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64
// CHECK9-NEXT: [[DIV21:%.*]] = sdiv i64 [[TMP21]], [[CONV20]] // CHECK9-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP21]], [[CONV18]]
// CHECK9-NEXT: [[MUL22:%.*]] = mul nsw i64 [[DIV21]], 1 // CHECK9-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL22]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]]
// CHECK9-NEXT: [[CONV23:%.*]] = trunc i64 [[ADD]] to i32 // CHECK9-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32
// CHECK9-NEXT: store i32 [[CONV23]], ptr [[I13]], align 4 // CHECK9-NEXT: store i32 [[CONV21]], ptr [[I11]], align 4
// CHECK9-NEXT: [[TMP23:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP23:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT: [[TMP24:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP24:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP25]], 0 // CHECK9-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP25]], 0
// CHECK9-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1 // CHECK9-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1
// CHECK9-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]] // CHECK9-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]]
// CHECK9-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64 // CHECK9-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64
// CHECK9-NEXT: [[DIV28:%.*]] = sdiv i64 [[TMP24]], [[CONV27]] // CHECK9-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP24]], [[CONV25]]
// CHECK9-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP26]], 0 // CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK9-NEXT: [[DIV30:%.*]] = sdiv i32 [[SUB29]], 1 // CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK9-NEXT: [[MUL31:%.*]] = mul nsw i32 1, [[DIV30]] // CHECK9-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]]
// CHECK9-NEXT: [[CONV32:%.*]] = sext i32 [[MUL31]] to i64 // CHECK9-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64
// CHECK9-NEXT: [[MUL33:%.*]] = mul nsw i64 [[DIV28]], [[CONV32]] // CHECK9-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]]
// CHECK9-NEXT: [[SUB34:%.*]] = sub nsw i64 [[TMP23]], [[MUL33]] // CHECK9-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP23]], [[MUL31]]
// CHECK9-NEXT: [[MUL35:%.*]] = mul nsw i64 [[SUB34]], 1 // CHECK9-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1
// CHECK9-NEXT: [[ADD36:%.*]] = add nsw i64 0, [[MUL35]] // CHECK9-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]]
// CHECK9-NEXT: [[CONV37:%.*]] = trunc i64 [[ADD36]] to i32 // CHECK9-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32
// CHECK9-NEXT: store i32 [[CONV37]], ptr [[J14]], align 4 // CHECK9-NEXT: store i32 [[CONV35]], ptr [[J12]], align 4
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, ptr [[I13]], align 4 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, ptr [[I11]], align 4
// CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP27]] to i64 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP27]] to i64
// CHECK9-NEXT: [[TMP28:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP1]] // CHECK9-NEXT: [[TMP28:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP1]]
// CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[TMP28]] // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[TMP28]]
// CHECK9-NEXT: [[TMP29:%.*]] = load i32, ptr [[J14]], align 4 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, ptr [[J12]], align 4
// CHECK9-NEXT: [[IDXPROM38:%.*]] = sext i32 [[TMP29]] to i64 // CHECK9-NEXT: [[IDXPROM36:%.*]] = sext i32 [[TMP29]] to i64
// CHECK9-NEXT: [[ARRAYIDX39:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i64 [[IDXPROM38]] // CHECK9-NEXT: [[ARRAYIDX37:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i64 [[IDXPROM36]]
// CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX39]], align 4 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX37]], align 4
// 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, ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: [[TMP30:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK9-NEXT: [[ADD40:%.*]] = add nsw i64 [[TMP30]], 1 // CHECK9-NEXT: [[ADD38:%.*]] = add nsw i64 [[TMP30]], 1
// CHECK9-NEXT: store i64 [[ADD40]], ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: store i64 [[ADD38]], ptr [[DOTOMP_IV]], align 8
// 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: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP32]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP32]])
// CHECK9-NEXT: br label [[OMP_PRECOND_END]] // CHECK9-NEXT: br label [[OMP_PRECOND_END]]
Show All 9 Lines
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store ptr [[A]], ptr [[TMP0]], align 8 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP0]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: store ptr [[A]], ptr [[TMP2]], align 8 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP1]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK9-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK9-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK9-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK9-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK9-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8 // CHECK9-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK9-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK9-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK9-NEXT: store ptr @.offload_sizes.4, ptr [[TMP11]], align 8 // CHECK9-NEXT: store ptr @.offload_sizes.4, ptr [[TMP9]], align 8
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK9-NEXT: store ptr @.offload_maptypes.5, ptr [[TMP12]], align 8 // CHECK9-NEXT: store ptr @.offload_maptypes.5, ptr [[TMP10]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK9-NEXT: store ptr null, ptr [[TMP13]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK9-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK9-NEXT: store i64 20, ptr [[TMP15]], align 8 // CHECK9-NEXT: store i64 20, ptr [[TMP13]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, ptr [[KERNEL_ARGS]]) // CHECK9-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, ptr [[KERNEL_ARGS]])
// CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP15]], 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(ptr [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68(ptr [[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
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68
▲ Show 20 Lines • Show All 209 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[M]], align 4 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP6]], ptr [[M_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP6]], ptr [[M_CASTED]], align 4
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[M_CASTED]], align 4 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[M_CASTED]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP0]], [[TMP1]] // CHECK11-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP0]], [[TMP1]]
// 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: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 40, i1 false) // CHECK11-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 40, i1 false)
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP11]], align 4
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP12]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP12]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP14]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP13]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store ptr null, ptr [[TMP16]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP16]], align 4
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP17]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP17]], align 4
// CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP19]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP18]], align 4
// CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: store ptr null, ptr [[TMP21]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP19]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP22]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP20]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP24]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP21]], align 4
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK11-NEXT: store ptr null, ptr [[TMP22]], align 4
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP23]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP24]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK11-NEXT: store i64 [[TMP10]], ptr [[TMP25]], align 4
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK11-NEXT: store ptr null, ptr [[TMP26]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP26]], align 4
// CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP27]], align 4 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP29]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK11-NEXT: store i32 [[TMP30]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store ptr null, ptr [[TMP31]], align 4 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, ptr [[M]], align 4
// CHECK11-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK11-NEXT: store i32 [[TMP31]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP32]], align 4 // CHECK11-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP32]], 0
// CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP34]], align 4
// CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK11-NEXT: store i64 [[TMP10]], ptr [[TMP36]], align 4
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK11-NEXT: store ptr null, ptr [[TMP37]], align 4
// CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP41:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP41]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = load i32, ptr [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP42]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP43:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP43]], 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: [[TMP44:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP44]], 0 // CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP33]], 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]], ptr [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: store i64 [[SUB7]], ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[TMP45:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: [[TMP34:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP45]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP34]], 1
// CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK11-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK11-NEXT: store i32 1, ptr [[TMP46]], align 4 // CHECK11-NEXT: store i32 1, ptr [[TMP35]], align 4
// CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK11-NEXT: store i32 5, ptr [[TMP47]], align 4 // CHECK11-NEXT: store i32 5, ptr [[TMP36]], align 4
// CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK11-NEXT: store ptr [[TMP38]], ptr [[TMP48]], align 4 // CHECK11-NEXT: store ptr [[TMP27]], ptr [[TMP37]], align 4
// CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK11-NEXT: store ptr [[TMP39]], ptr [[TMP49]], align 4 // CHECK11-NEXT: store ptr [[TMP28]], ptr [[TMP38]], align 4
// CHECK11-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK11-NEXT: store ptr [[TMP40]], ptr [[TMP50]], align 4 // CHECK11-NEXT: store ptr [[TMP29]], ptr [[TMP39]], align 4
// CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK11-NEXT: store ptr @.offload_maptypes, ptr [[TMP51]], align 4 // CHECK11-NEXT: store ptr @.offload_maptypes, ptr [[TMP40]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK11-NEXT: store ptr null, ptr [[TMP52]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP41]], align 4
// CHECK11-NEXT: [[TMP53:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK11-NEXT: store ptr null, ptr [[TMP53]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP42]], align 4
// CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK11-NEXT: store i64 [[ADD]], ptr [[TMP54]], align 8 // CHECK11-NEXT: store i64 [[ADD]], ptr [[TMP43]], align 8
// CHECK11-NEXT: [[TMP55:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, ptr [[KERNEL_ARGS]]) // CHECK11-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, ptr [[KERNEL_ARGS]])
// CHECK11-NEXT: [[TMP56:%.*]] = icmp ne i32 [[TMP55]], 0 // CHECK11-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK11-NEXT: br i1 [[TMP56]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP45]], 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]], ptr [[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]], ptr [[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: [[TMP57:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK11-NEXT: [[TMP46:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef [[TMP57]]) // CHECK11-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef [[TMP46]])
// CHECK11-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP58:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(ptr [[TMP58]]) // CHECK11-NEXT: call void @llvm.stackrestore(ptr [[TMP47]])
// CHECK11-NEXT: [[TMP59:%.*]] = load i32, ptr [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP48:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP59]] // CHECK11-NEXT: ret i32 [[TMP48]]
// //
// //
// 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 noundef [[N:%.*]], i32 noundef [[M:%.*]], i32 noundef [[VLA:%.*]], i32 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[M:%.*]], i32 noundef [[VLA:%.*]], i32 noundef [[VLA1:%.*]], ptr noundef 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 301 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], 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: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store ptr [[A]], ptr [[TMP0]], align 4 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP0]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: store ptr [[A]], ptr [[TMP2]], align 4 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP1]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK11-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK11-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK11-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK11-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK11-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 // CHECK11-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK11-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK11-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK11-NEXT: store ptr @.offload_sizes.4, ptr [[TMP11]], align 4 // CHECK11-NEXT: store ptr @.offload_sizes.4, ptr [[TMP9]], align 4
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK11-NEXT: store ptr @.offload_maptypes.5, ptr [[TMP12]], align 4 // CHECK11-NEXT: store ptr @.offload_maptypes.5, ptr [[TMP10]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK11-NEXT: store ptr null, ptr [[TMP13]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK11-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK11-NEXT: store i64 20, ptr [[TMP15]], align 8 // CHECK11-NEXT: store i64 20, ptr [[TMP13]], align 8
// CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, ptr [[KERNEL_ARGS]]) // CHECK11-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, ptr [[KERNEL_ARGS]])
// CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK11-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP15]], 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(ptr [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68(ptr [[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
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68
▲ Show 20 Lines • Show All 165 Lines • Show Last 20 Lines

clang/test/OpenMP/target_teams_distribute_parallel_for_if_codegen.cpp

Show First 20 Lines • Show All 493 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store i8 [[FROMBOOL2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1 // CHECK1-NEXT: store i8 [[FROMBOOL2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1
// CHECK1-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK1-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1 // CHECK1-NEXT: [[TMP14:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1
// CHECK1-NEXT: [[TOBOOL3:%.*]] = trunc i8 [[TMP14]] to i1 // CHECK1-NEXT: [[TOBOOL3:%.*]] = trunc i8 [[TMP14]] to i1
// CHECK1-NEXT: br i1 [[TOBOOL3]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] // CHECK1-NEXT: br i1 [[TOBOOL3]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]]
// CHECK1: omp_if.then: // CHECK1: omp_if.then:
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store i64 [[TMP13]], ptr [[TMP15]], align 8 // CHECK1-NEXT: store i64 [[TMP13]], ptr [[TMP15]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store i64 [[TMP13]], ptr [[TMP17]], align 8 // CHECK1-NEXT: store i64 [[TMP13]], ptr [[TMP16]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP19]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP17]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP22:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1 // CHECK1-NEXT: [[TMP20:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1
// CHECK1-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP22]] to i1 // CHECK1-NEXT: [[TOBOOL4:%.*]] = trunc i8 [[TMP20]] to i1
// CHECK1-NEXT: [[TMP23:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1 // CHECK1-NEXT: [[TMP21:%.*]] = select i1 [[TOBOOL4]], i32 0, i32 1
// CHECK1-NEXT: [[KERNEL_ARGS6:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS6:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 0 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP24]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP22]], align 4
// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 1 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP25]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP23]], align 4
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 2 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP20]], ptr [[TMP26]], align 8 // CHECK1-NEXT: store ptr [[TMP18]], ptr [[TMP24]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 3 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP21]], ptr [[TMP27]], align 8 // CHECK1-NEXT: store ptr [[TMP19]], ptr [[TMP25]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 4 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP28]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP26]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 5 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP29]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP27]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 6 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP30]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP28]], align 8
// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 7 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP31]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP29]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 8 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS6]], i32 0, i32 8
// CHECK1-NEXT: store i64 100, ptr [[TMP32]], align 8 // CHECK1-NEXT: store i64 100, ptr [[TMP30]], align 8
// CHECK1-NEXT: [[TMP33:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 [[TMP23]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l90.region_id, ptr [[KERNEL_ARGS6]]) // CHECK1-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 [[TMP21]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l90.region_id, ptr [[KERNEL_ARGS6]])
// CHECK1-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0 // CHECK1-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0
// CHECK1-NEXT: br i1 [[TMP34]], label [[OMP_OFFLOAD_FAILED7:%.*]], label [[OMP_OFFLOAD_CONT8:%.*]] // CHECK1-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED7:%.*]], label [[OMP_OFFLOAD_CONT8:%.*]]
// CHECK1: omp_offload.failed7: // CHECK1: omp_offload.failed7:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l90(i64 [[TMP13]]) #[[ATTR2]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l90(i64 [[TMP13]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT8]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT8]]
// CHECK1: omp_offload.cont8: // CHECK1: omp_offload.cont8:
// 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]+}}_main_l90(i64 [[TMP13]]) #[[ATTR2]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l90(i64 [[TMP13]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_IF_END]] // CHECK1-NEXT: br label [[OMP_IF_END]]
// CHECK1: omp_if.end: // CHECK1: omp_if.end:
// CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr @Arg, align 4 // CHECK1-NEXT: [[TMP33:%.*]] = load i32, ptr @Arg, align 4
// CHECK1-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiEiT_(i32 noundef [[TMP35]]) // CHECK1-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiEiT_(i32 noundef [[TMP33]])
// CHECK1-NEXT: ret i32 [[CALL]] // CHECK1-NEXT: ret i32 [[CALL]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l76
// CHECK1-SAME: () #[[ATTR1]] { // CHECK1-SAME: () #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 0, ptr @.omp_outlined..4) // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 0, ptr @.omp_outlined..4)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
▲ Show 20 Lines • Show All 480 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store i8 [[FROMBOOL]], ptr [[DOTCAPTURE_EXPR_]], align 1 // CHECK1-NEXT: store i8 [[FROMBOOL]], ptr [[DOTCAPTURE_EXPR_]], align 1
// CHECK1-NEXT: [[TMP12:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1 // CHECK1-NEXT: [[TMP12:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1
// CHECK1-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP12]] to i1 // CHECK1-NEXT: [[TOBOOL1:%.*]] = trunc i8 [[TMP12]] to i1
// CHECK1-NEXT: [[FROMBOOL2:%.*]] = zext i1 [[TOBOOL1]] to i8 // CHECK1-NEXT: [[FROMBOOL2:%.*]] = zext i1 [[TOBOOL1]] to i8
// CHECK1-NEXT: store i8 [[FROMBOOL2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1 // CHECK1-NEXT: store i8 [[FROMBOOL2]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1
// CHECK1-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8 // CHECK1-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store i64 [[TMP13]], ptr [[TMP14]], align 8 // CHECK1-NEXT: store i64 [[TMP13]], ptr [[TMP14]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store i64 [[TMP13]], ptr [[TMP16]], align 8 // CHECK1-NEXT: store i64 [[TMP13]], ptr [[TMP15]], align 8
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP18]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP16]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP21:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1 // CHECK1-NEXT: [[TMP19:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1
// CHECK1-NEXT: [[TOBOOL3:%.*]] = trunc i8 [[TMP21]] to i1 // CHECK1-NEXT: [[TOBOOL3:%.*]] = trunc i8 [[TMP19]] to i1
// CHECK1-NEXT: [[TMP22:%.*]] = select i1 [[TOBOOL3]], i32 0, i32 1 // CHECK1-NEXT: [[TMP20:%.*]] = select i1 [[TOBOOL3]], i32 0, i32 1
// CHECK1-NEXT: [[KERNEL_ARGS5:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS5:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 0 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP23]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP21]], align 4
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 1 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP24]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP22]], align 4
// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 2 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP19]], ptr [[TMP25]], align 8 // CHECK1-NEXT: store ptr [[TMP17]], ptr [[TMP23]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 3 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP20]], ptr [[TMP26]], align 8 // CHECK1-NEXT: store ptr [[TMP18]], ptr [[TMP24]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 4 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.16, ptr [[TMP27]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.16, ptr [[TMP25]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 5 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.17, ptr [[TMP28]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.17, ptr [[TMP26]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 6 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP29]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP27]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 7 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP30]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP28]], align 8
// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 8 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 8
// CHECK1-NEXT: store i64 100, ptr [[TMP31]], align 8 // CHECK1-NEXT: store i64 100, ptr [[TMP29]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 [[TMP22]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l68.region_id, ptr [[KERNEL_ARGS5]]) // CHECK1-NEXT: [[TMP30:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 [[TMP20]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l68.region_id, ptr [[KERNEL_ARGS5]])
// CHECK1-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0 // CHECK1-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0
// CHECK1-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]] // CHECK1-NEXT: br i1 [[TMP31]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]]
// CHECK1: omp_offload.failed6: // CHECK1: omp_offload.failed6:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l68(i64 [[TMP13]]) #[[ATTR2]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l68(i64 [[TMP13]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT7]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT7]]
// CHECK1: omp_offload.cont7: // CHECK1: omp_offload.cont7:
// CHECK1-NEXT: ret i32 0 // CHECK1-NEXT: ret i32 0
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l60 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiEiT__l60
▲ Show 20 Lines • Show All 448 Lines • Show Last 20 Lines

clang/test/OpenMP/target_teams_distribute_parallel_for_order_codegen.cpp

Show First 20 Lines • Show All 161 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 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 99, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP6]], ptr [[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: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4:![0-9]+]]
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// 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: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP4]]
// 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: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![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(ptr @[[GLOB1]], i32 [[TMP3]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP3]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK1-SAME: () #[[ATTR3:[0-9]+]] section ".text.startup" { // CHECK1-SAME: () #[[ATTR3:[0-9]+]] section ".text.startup" {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: call void @__tgt_register_requires(i64 1) // CHECK1-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //

clang/test/OpenMP/target_teams_distribute_parallel_for_reduction_codegen.cpp

Show First 20 Lines • Show All 95 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK1-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 8 // CHECK1-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 2, ptr [[TMP15]], align 8 // CHECK1-NEXT: store i64 2, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB4:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l66.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB4:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l66.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP15]], 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]+}}_main_l66(ptr @_ZZ4mainE5sivar) #[[ATTR2:[0-9]+]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l66(ptr @_ZZ4mainE5sivar) #[[ATTR2:[0-9]+]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v() // CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v()
// CHECK1-NEXT: ret i32 [[CALL]] // CHECK1-NEXT: ret i32 [[CALL]]
// //
// //
▲ Show 20 Lines • Show All 67 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[SIVAR1]], ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr [[SIVAR1]], ptr [[TMP14]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK1-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK1-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4 // CHECK1-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 68 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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(ptr @[[GLOB1]], i32 [[TMP4]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP4]])
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[SIVAR2]], ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr [[SIVAR2]], ptr [[TMP14]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[SIVAR2]], align 4 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[SIVAR2]], align 4
// CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK1-NEXT: store i32 [[ADD6]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD6]], ptr [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[SIVAR2]], align 4 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[SIVAR2]], align 4
// CHECK1-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4 // CHECK1-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v // CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK1-SAME: () #[[ATTR5:[0-9]+]] comdat { // CHECK1-SAME: () #[[ATTR5:[0-9]+]] comdat {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[T_VAR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4 // CHECK1-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32 0, ptr [[T_VAR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[T_VAR]], align 4
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i64 8, i1 false) // CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i64 8, i1 false)
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[T_VAR]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[T_VAR]], ptr [[TMP1]], align 8 // CHECK1-NEXT: store ptr [[T_VAR]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[T_VAR]], ptr [[TMP3]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr null, ptr [[TMP5]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP9]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 8 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.7, ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.7, ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP13]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP15]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 2, ptr [[TMP16]], align 8 // CHECK1-NEXT: store i64 2, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB4]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB4]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP18]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP15]], 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]+}}__Z5tmainIiET_v_l32(ptr [[T_VAR]]) #[[ATTR2]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32(ptr [[T_VAR]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: ret i32 0 // CHECK1-NEXT: ret i32 0
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32
▲ Show 20 Lines • Show All 66 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[T_VAR1]], ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr [[T_VAR1]], ptr [[TMP14]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.6, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.6, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK1-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK1-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4 // CHECK1-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..4 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[T_VAR:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[T_VAR:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 68 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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(ptr @[[GLOB1]], i32 [[TMP4]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP4]])
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[T_VAR2]], ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr [[T_VAR2]], ptr [[TMP14]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.5, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.5, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[T_VAR2]], align 4 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[T_VAR2]], align 4
// CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK1-NEXT: store i32 [[ADD6]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD6]], ptr [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[T_VAR2]], align 4 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[T_VAR2]], align 4
// CHECK1-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4 // CHECK1-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.5 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.5
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.6 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.6
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK1-SAME: () #[[ATTR7:[0-9]+]] { // CHECK1-SAME: () #[[ATTR7:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: call void @__tgt_register_requires(i64 1) // CHECK1-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@main // CHECK3-LABEL: define {{[^@]+}}@main
// CHECK3-SAME: () #[[ATTR0:[0-9]+]] { // CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK3-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 4 // CHECK3-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP2]], align 4 // CHECK3-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 2, ptr [[TMP15]], align 8 // CHECK3-NEXT: store i64 2, ptr [[TMP13]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB4:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l66.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB4:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l66.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP15]], 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]+}}_main_l66(ptr @_ZZ4mainE5sivar) #[[ATTR2:[0-9]+]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l66(ptr @_ZZ4mainE5sivar) #[[ATTR2:[0-9]+]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v() // CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v()
// CHECK3-NEXT: ret i32 [[CALL]] // CHECK3-NEXT: ret i32 [[CALL]]
// //
// //
▲ Show 20 Lines • Show All 65 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[SIVAR1]], ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr [[SIVAR1]], ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3:[0-9]+]], i32 [[TMP2]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: [[TMP13:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3:[0-9]+]], i32 [[TMP2]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK3-NEXT: switch i32 [[TMP13]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK3-NEXT: ] // CHECK3-NEXT: ]
// CHECK3: .omp.reduction.case1: // CHECK3: .omp.reduction.case1:
// CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]] // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK3-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.case2: // CHECK3: .omp.reduction.case2:
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4 // CHECK3-NEXT: [[TMP17:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP16]] monotonic, align 4
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.default: // CHECK3: .omp.reduction.default:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..1 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR1]] { // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
▲ Show 20 Lines • Show All 66 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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(ptr @[[GLOB1]], i32 [[TMP4]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP4]])
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[SIVAR1]], ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr [[SIVAR1]], ptr [[TMP14]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK3-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK3-NEXT: ] // CHECK3-NEXT: ]
// CHECK3: .omp.reduction.case1: // CHECK3: .omp.reduction.case1:
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.case2: // CHECK3: .omp.reduction.case2:
// CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK3-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4 // CHECK3-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.default: // CHECK3: .omp.reduction.default:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { // CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2 // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@_Z5tmainIiET_v // CHECK3-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK3-SAME: () #[[ATTR5:[0-9]+]] comdat { // CHECK3-SAME: () #[[ATTR5:[0-9]+]] comdat {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[T_VAR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4 // CHECK3-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 0, ptr [[T_VAR]], align 4 // CHECK3-NEXT: store i32 0, ptr [[T_VAR]], align 4
// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i32 8, i1 false) // CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i32 8, i1 false)
// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR]], ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR]], ptr [[TMP1]], align 4 // CHECK3-NEXT: store ptr [[T_VAR]], ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR]], ptr [[TMP3]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP5]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, ptr [[TMP9]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 4 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes.7, ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes.7, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP13]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP15]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 2, ptr [[TMP16]], align 8 // CHECK3-NEXT: store i64 2, ptr [[TMP13]], align 8
// CHECK3-NEXT: [[TMP17:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB4]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB4]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP18]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP15]], 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]+}}__Z5tmainIiET_v_l32(ptr [[T_VAR]]) #[[ATTR2]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32(ptr [[T_VAR]]) #[[ATTR2]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: ret i32 0 // CHECK3-NEXT: ret i32 0
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32
▲ Show 20 Lines • Show All 64 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR1]], ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr [[T_VAR1]], ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.6, ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: [[TMP13:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.6, ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK3-NEXT: switch i32 [[TMP13]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK3-NEXT: ] // CHECK3-NEXT: ]
// CHECK3: .omp.reduction.case1: // CHECK3: .omp.reduction.case1:
// CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]] // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK3-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.case2: // CHECK3: .omp.reduction.case2:
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4 // CHECK3-NEXT: [[TMP17:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP16]] monotonic, align 4
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.default: // CHECK3: .omp.reduction.default:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..4 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[T_VAR:%.*]]) #[[ATTR1]] { // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[T_VAR:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
▲ Show 20 Lines • Show All 66 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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(ptr @[[GLOB1]], i32 [[TMP4]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP4]])
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR1]], ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr [[T_VAR1]], ptr [[TMP14]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.5, ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.5, ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK3-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK3-NEXT: ] // CHECK3-NEXT: ]
// CHECK3: .omp.reduction.case1: // CHECK3: .omp.reduction.case1:
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.case2: // CHECK3: .omp.reduction.case2:
// CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK3-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4 // CHECK3-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.default: // CHECK3: .omp.reduction.default:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.5 // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.5
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.6 // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.6
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK3-SAME: () #[[ATTR7:[0-9]+]] { // CHECK3-SAME: () #[[ATTR7:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: call void @__tgt_register_requires(i64 1) // CHECK3-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
▲ Show 20 Lines • Show All 79 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK5: omp.inner.for.end: // CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit: // CHECK5: omp.loop.exit:
// CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK5-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK5-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK5-NEXT: store ptr [[SIVAR1]], ptr [[TMP14]], align 8 // CHECK5-NEXT: store ptr [[SIVAR1]], ptr [[TMP14]], align 8
// CHECK5-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK5-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK5-NEXT: ] // CHECK5-NEXT: ]
// CHECK5: .omp.reduction.case1: // CHECK5: .omp.reduction.case1:
// CHECK5-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK5-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK5-NEXT: [[TMP19:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK5-NEXT: [[TMP17:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK5-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4 // CHECK5-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4
// CHECK5-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.case2: // CHECK5: .omp.reduction.case2:
// CHECK5-NEXT: [[TMP20:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK5-NEXT: [[TMP18:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK5-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4 // CHECK5-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.default: // CHECK5: .omp.reduction.default:
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..1 // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR2]] { // CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR2]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
▲ Show 20 Lines • Show All 72 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]]
// CHECK5: omp.inner.for.end: // CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit: // CHECK5: omp.loop.exit:
// CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP4]]) // CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP4]])
// CHECK5-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK5-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK5-NEXT: store ptr [[SIVAR2]], ptr [[TMP15]], align 8 // CHECK5-NEXT: store ptr [[SIVAR2]], ptr [[TMP15]], align 8
// CHECK5-NEXT: [[TMP18:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: [[TMP16:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: switch i32 [[TMP18]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK5-NEXT: switch i32 [[TMP16]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK5-NEXT: ] // CHECK5-NEXT: ]
// CHECK5: .omp.reduction.case1: // CHECK5: .omp.reduction.case1:
// CHECK5-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK5-NEXT: [[TMP17:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK5-NEXT: [[TMP20:%.*]] = load i32, ptr [[SIVAR2]], align 4 // CHECK5-NEXT: [[TMP18:%.*]] = load i32, ptr [[SIVAR2]], align 4
// CHECK5-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] // CHECK5-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP17]], [[TMP18]]
// CHECK5-NEXT: store i32 [[ADD6]], ptr [[TMP0]], align 4 // CHECK5-NEXT: store i32 [[ADD6]], ptr [[TMP0]], align 4
// CHECK5-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.case2: // CHECK5: .omp.reduction.case2:
// CHECK5-NEXT: [[TMP21:%.*]] = load i32, ptr [[SIVAR2]], align 4 // CHECK5-NEXT: [[TMP19:%.*]] = load i32, ptr [[SIVAR2]], align 4
// CHECK5-NEXT: [[TMP22:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP21]] monotonic, align 4 // CHECK5-NEXT: [[TMP20:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP19]] monotonic, align 4
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.default: // CHECK5: .omp.reduction.default:
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4:[0-9]+]] { // CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK5-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK5-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK5-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK5-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2 // CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2
// CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4]] { // CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK5-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK5-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK5-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK5-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK5-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK5-SAME: () #[[ATTR6:[0-9]+]] { // CHECK5-SAME: () #[[ATTR6:[0-9]+]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: call void @__tgt_register_requires(i64 1) // CHECK5-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //

clang/test/OpenMP/target_teams_distribute_parallel_for_reduction_task_codegen.cpp

Show First 20 Lines • Show All 127 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP19:%.*]] = sub i64 [[TMP17]], [[TMP18]] // CHECK1-NEXT: [[TMP19:%.*]] = sub i64 [[TMP17]], [[TMP18]]
// CHECK1-NEXT: [[TMP20:%.*]] = sdiv exact i64 [[TMP19]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP20:%.*]] = sdiv exact i64 [[TMP19]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[VLA]], i64 [[TMP20]] // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[VLA]], i64 [[TMP20]]
// CHECK1-NEXT: store ptr [[_TMP5]], ptr [[TMP]], align 8 // CHECK1-NEXT: store ptr [[_TMP5]], ptr [[TMP]], align 8
// CHECK1-NEXT: store ptr [[TMP21]], ptr [[_TMP5]], align 8 // CHECK1-NEXT: store ptr [[TMP21]], ptr [[_TMP5]], align 8
// CHECK1-NEXT: [[DOTRD_INPUT_GEP_:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], ptr [[DOTRD_INPUT_]], i64 0, i64 0 // CHECK1-NEXT: [[DOTRD_INPUT_GEP_:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], ptr [[DOTRD_INPUT_]], i64 0, i64 0
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T:%.*]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 0 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T:%.*]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP22]], align 8 // CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP22]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 1 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP24]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP23]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 2 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 2
// CHECK1-NEXT: store i64 4, ptr [[TMP26]], align 8 // CHECK1-NEXT: store i64 4, ptr [[TMP24]], align 8
// CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 3 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 3
// CHECK1-NEXT: store ptr @.red_init., ptr [[TMP27]], align 8 // CHECK1-NEXT: store ptr @.red_init., ptr [[TMP25]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 4 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 4
// CHECK1-NEXT: store ptr null, ptr [[TMP28]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP26]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 5 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.red_comb., ptr [[TMP29]], align 8 // CHECK1-NEXT: store ptr @.red_comb., ptr [[TMP27]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 6 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 6
// CHECK1-NEXT: call void @llvm.memset.p0.i64(ptr align 8 [[TMP30]], i8 0, i64 4, i1 false) // CHECK1-NEXT: call void @llvm.memset.p0.i64(ptr align 8 [[TMP28]], i8 0, i64 4, i1 false)
// CHECK1-NEXT: [[DOTRD_INPUT_GEP_6:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], ptr [[DOTRD_INPUT_]], i64 0, i64 1 // CHECK1-NEXT: [[DOTRD_INPUT_GEP_6:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], ptr [[DOTRD_INPUT_]], i64 0, i64 1
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 0 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 0
// CHECK1-NEXT: [[TMP33:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8 // CHECK1-NEXT: [[TMP30:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8
// CHECK1-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds ptr, ptr [[TMP33]], i64 0 // CHECK1-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds ptr, ptr [[TMP30]], i64 0
// CHECK1-NEXT: [[TMP34:%.*]] = load ptr, ptr [[ARRAYIDX7]], align 8 // CHECK1-NEXT: [[TMP31:%.*]] = load ptr, ptr [[ARRAYIDX7]], align 8
// CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i8, ptr [[TMP34]], i64 0 // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i8, ptr [[TMP31]], i64 0
// CHECK1-NEXT: [[TMP35:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 // CHECK1-NEXT: [[TMP33:%.*]] = sext i32 [[TMP32]] to i64
// CHECK1-NEXT: [[LB_ADD_LEN9:%.*]] = add nsw i64 -1, [[TMP36]] // CHECK1-NEXT: [[LB_ADD_LEN9:%.*]] = add nsw i64 -1, [[TMP33]]
// CHECK1-NEXT: [[TMP37:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8 // CHECK1-NEXT: [[TMP34:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8
// CHECK1-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds ptr, ptr [[TMP37]], i64 9 // CHECK1-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds ptr, ptr [[TMP34]], i64 9
// CHECK1-NEXT: [[TMP38:%.*]] = load ptr, ptr [[ARRAYIDX10]], align 8 // CHECK1-NEXT: [[TMP35:%.*]] = load ptr, ptr [[ARRAYIDX10]], align 8
// CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds i8, ptr [[TMP38]], i64 [[LB_ADD_LEN9]] // CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds i8, ptr [[TMP35]], i64 [[LB_ADD_LEN9]]
// CHECK1-NEXT: store ptr [[VLA]], ptr [[TMP32]], align 8 // CHECK1-NEXT: store ptr [[VLA]], ptr [[TMP29]], align 8
// CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 1 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[ARRAYIDX8]], ptr [[TMP39]], align 8 // CHECK1-NEXT: store ptr [[ARRAYIDX8]], ptr [[TMP36]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = ptrtoint ptr [[ARRAYIDX11]] to i64 // CHECK1-NEXT: [[TMP37:%.*]] = ptrtoint ptr [[ARRAYIDX11]] to i64
// CHECK1-NEXT: [[TMP41:%.*]] = ptrtoint ptr [[ARRAYIDX8]] to i64 // CHECK1-NEXT: [[TMP38:%.*]] = ptrtoint ptr [[ARRAYIDX8]] to i64
// CHECK1-NEXT: [[TMP42:%.*]] = sub i64 [[TMP40]], [[TMP41]] // CHECK1-NEXT: [[TMP39:%.*]] = sub i64 [[TMP37]], [[TMP38]]
// CHECK1-NEXT: [[TMP43:%.*]] = sdiv exact i64 [[TMP42]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP40:%.*]] = sdiv exact i64 [[TMP39]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP44:%.*]] = add nuw i64 [[TMP43]], 1 // CHECK1-NEXT: [[TMP41:%.*]] = add nuw i64 [[TMP40]], 1
// CHECK1-NEXT: [[TMP45:%.*]] = mul nuw i64 [[TMP44]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP42:%.*]] = mul nuw i64 [[TMP41]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 2 // CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 2
// CHECK1-NEXT: store i64 [[TMP45]], ptr [[TMP46]], align 8 // CHECK1-NEXT: store i64 [[TMP42]], ptr [[TMP43]], align 8
// CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 3 // CHECK1-NEXT: [[TMP44:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 3
// CHECK1-NEXT: store ptr @.red_init..1, ptr [[TMP47]], align 8 // CHECK1-NEXT: store ptr @.red_init..1, ptr [[TMP44]], align 8
// CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 4 // CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 4
// CHECK1-NEXT: store ptr null, ptr [[TMP48]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP45]], align 8
// CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 5 // CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.red_comb..2, ptr [[TMP49]], align 8 // CHECK1-NEXT: store ptr @.red_comb..2, ptr [[TMP46]], align 8
// CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 6 // CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], ptr [[DOTRD_INPUT_GEP_6]], i32 0, i32 6
// CHECK1-NEXT: store i32 1, ptr [[TMP50]], align 8 // CHECK1-NEXT: store i32 1, ptr [[TMP47]], align 8
// CHECK1-NEXT: [[TMP51:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP48:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP52:%.*]] = load i32, ptr [[TMP51]], align 4 // CHECK1-NEXT: [[TMP49:%.*]] = load i32, ptr [[TMP48]], align 4
// CHECK1-NEXT: [[TMP54:%.*]] = call ptr @__kmpc_taskred_modifier_init(ptr @[[GLOB1]], i32 [[TMP52]], i32 1, i32 2, ptr [[DOTRD_INPUT_]]) // CHECK1-NEXT: [[TMP50:%.*]] = call ptr @__kmpc_taskred_modifier_init(ptr @[[GLOB1]], i32 [[TMP49]], i32 1, i32 2, ptr [[DOTRD_INPUT_]])
// CHECK1-NEXT: store ptr [[TMP54]], ptr [[DOTTASK_RED_]], align 8 // CHECK1-NEXT: store ptr [[TMP50]], ptr [[DOTTASK_RED_]], align 8
// CHECK1-NEXT: store i64 0, ptr [[DOTOMP_COMB_LB]], align 8 // CHECK1-NEXT: store i64 0, ptr [[DOTOMP_COMB_LB]], align 8
// CHECK1-NEXT: store i64 9, ptr [[DOTOMP_COMB_UB]], align 8 // CHECK1-NEXT: store i64 9, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK1-NEXT: store i64 1, ptr [[DOTOMP_STRIDE]], align 8 // CHECK1-NEXT: store i64 1, ptr [[DOTOMP_STRIDE]], align 8
// CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP55:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP51:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP56:%.*]] = load i32, ptr [[TMP55]], align 4 // CHECK1-NEXT: [[TMP52:%.*]] = load i32, ptr [[TMP51]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB2:[0-9]+]], i32 [[TMP56]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1) // CHECK1-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB2:[0-9]+]], i32 [[TMP52]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK1-NEXT: [[TMP57:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8 // CHECK1-NEXT: [[TMP53:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i64 [[TMP57]], 9 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i64 [[TMP53]], 9
// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1: cond.true: // CHECK1: cond.true:
// CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1-NEXT: br label [[COND_END:%.*]]
// CHECK1: cond.false: // CHECK1: cond.false:
// CHECK1-NEXT: [[TMP58:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8 // CHECK1-NEXT: [[TMP54:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK1-NEXT: br label [[COND_END]] // CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end: // CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i64 [ 9, [[COND_TRUE]] ], [ [[TMP58]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i64 [ 9, [[COND_TRUE]] ], [ [[TMP54]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i64 [[COND]], ptr [[DOTOMP_COMB_UB]], align 8 // CHECK1-NEXT: store i64 [[COND]], ptr [[DOTOMP_COMB_UB]], align 8
// CHECK1-NEXT: [[TMP59:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8 // CHECK1-NEXT: [[TMP55:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
// CHECK1-NEXT: store i64 [[TMP59]], ptr [[DOTOMP_IV]], align 8 // CHECK1-NEXT: store i64 [[TMP55]], ptr [[DOTOMP_IV]], align 8
// 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: [[TMP60:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK1-NEXT: [[TMP56:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK1-NEXT: [[TMP61:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8 // CHECK1-NEXT: [[TMP57:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK1-NEXT: [[CMP13:%.*]] = icmp sle i64 [[TMP60]], [[TMP61]] // CHECK1-NEXT: [[CMP13:%.*]] = icmp sle i64 [[TMP56]], [[TMP57]]
// CHECK1-NEXT: br i1 [[CMP13]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1-NEXT: br i1 [[CMP13]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body: // CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP62:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8 // CHECK1-NEXT: [[TMP58:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
// CHECK1-NEXT: [[TMP63:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8 // CHECK1-NEXT: [[TMP59:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK1-NEXT: [[TMP64:%.*]] = load ptr, ptr [[TMP]], align 8 // CHECK1-NEXT: [[TMP60:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 4, ptr @.omp_outlined..3, i64 [[TMP62]], i64 [[TMP63]], ptr [[ARGC1]], ptr [[TMP64]]) // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB1]], i32 4, ptr @.omp_outlined..3, i64 [[TMP58]], i64 [[TMP59]], ptr [[ARGC1]], ptr [[TMP60]])
// 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: [[TMP65:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK1-NEXT: [[TMP61:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK1-NEXT: [[TMP66:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8 // CHECK1-NEXT: [[TMP62:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP65]], [[TMP66]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP61]], [[TMP62]]
// CHECK1-NEXT: store i64 [[ADD]], ptr [[DOTOMP_IV]], align 8 // CHECK1-NEXT: store i64 [[ADD]], ptr [[DOTOMP_IV]], align 8
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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: [[TMP67:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP63:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP68:%.*]] = load i32, ptr [[TMP67]], align 4 // CHECK1-NEXT: [[TMP64:%.*]] = load i32, ptr [[TMP63]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP68]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP64]])
// CHECK1-NEXT: [[TMP69:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP65:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP70:%.*]] = load i32, ptr [[TMP69]], align 4 // CHECK1-NEXT: [[TMP66:%.*]] = load i32, ptr [[TMP65]], align 4
// CHECK1-NEXT: call void @__kmpc_task_reduction_modifier_fini(ptr @[[GLOB1]], i32 [[TMP70]], i32 1) // CHECK1-NEXT: call void @__kmpc_task_reduction_modifier_fini(ptr @[[GLOB1]], i32 [[TMP66]], i32 1)
// CHECK1-NEXT: [[TMP71:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP67:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP71]], align 8 // CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP67]], align 8
// CHECK1-NEXT: [[TMP73:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 // CHECK1-NEXT: [[TMP68:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT: store ptr [[VLA]], ptr [[TMP73]], align 8 // CHECK1-NEXT: store ptr [[VLA]], ptr [[TMP68]], align 8
// CHECK1-NEXT: [[TMP74:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 // CHECK1-NEXT: [[TMP69:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK1-NEXT: [[TMP75:%.*]] = inttoptr i64 [[TMP11]] to ptr // CHECK1-NEXT: [[TMP70:%.*]] = inttoptr i64 [[TMP11]] to ptr
// CHECK1-NEXT: store ptr [[TMP75]], ptr [[TMP74]], align 8 // CHECK1-NEXT: store ptr [[TMP70]], ptr [[TMP69]], align 8
// CHECK1-NEXT: [[TMP76:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP71:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP77:%.*]] = load i32, ptr [[TMP76]], align 4 // CHECK1-NEXT: [[TMP72:%.*]] = load i32, ptr [[TMP71]], align 4
// CHECK1-NEXT: [[TMP79:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB4:[0-9]+]], i32 [[TMP77]], i32 2, i64 24, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.9, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP73:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB4:[0-9]+]], i32 [[TMP72]], i32 2, i64 24, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.9, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP79]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP73]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP80:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP74:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP81:%.*]] = load i32, ptr [[ARGC1]], align 4 // CHECK1-NEXT: [[TMP75:%.*]] = load i32, ptr [[ARGC1]], align 4
// CHECK1-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP80]], [[TMP81]] // CHECK1-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP74]], [[TMP75]]
// CHECK1-NEXT: store i32 [[ADD14]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD14]], ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP82:%.*]] = getelementptr i8, ptr [[ARRAYIDX2]], i64 [[TMP11]] // CHECK1-NEXT: [[TMP76:%.*]] = getelementptr i8, ptr [[ARRAYIDX2]], i64 [[TMP11]]
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX2]], [[TMP82]] // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX2]], [[TMP76]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE21:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE21:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1: omp.arraycpy.body: // CHECK1: omp.arraycpy.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST15:%.*]] = phi ptr [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT19:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST15:%.*]] = phi ptr [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT19:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[TMP83:%.*]] = load i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST15]], align 1 // CHECK1-NEXT: [[TMP77:%.*]] = load i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST15]], align 1
// CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP83]] to i32 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP77]] to i32
// CHECK1-NEXT: [[TMP84:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 // CHECK1-NEXT: [[TMP78:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
// CHECK1-NEXT: [[CONV16:%.*]] = sext i8 [[TMP84]] to i32 // CHECK1-NEXT: [[CONV16:%.*]] = sext i8 [[TMP78]] to i32
// CHECK1-NEXT: [[ADD17:%.*]] = add nsw i32 [[CONV]], [[CONV16]] // CHECK1-NEXT: [[ADD17:%.*]] = add nsw i32 [[CONV]], [[CONV16]]
// CHECK1-NEXT: [[CONV18:%.*]] = trunc i32 [[ADD17]] to i8 // CHECK1-NEXT: [[CONV18:%.*]] = trunc i32 [[ADD17]] to i8
// CHECK1-NEXT: store i8 [[CONV18]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST15]], align 1 // CHECK1-NEXT: store i8 [[CONV18]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST15]], align 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT19]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST15]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT19]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST15]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE20:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT19]], [[TMP82]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE20:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT19]], [[TMP76]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE20]], label [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_BODY]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE20]], label [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1: omp.arraycpy.done21: // CHECK1: omp.arraycpy.done21:
// CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB4]], i32 [[TMP77]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB4]], i32 [[TMP72]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP85:%.*]] = load i32, ptr [[ARGC1]], align 4 // CHECK1-NEXT: [[TMP79:%.*]] = load i32, ptr [[ARGC1]], align 4
// CHECK1-NEXT: [[TMP86:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP85]] monotonic, align 4 // CHECK1-NEXT: [[TMP80:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP79]] monotonic, align 4
// CHECK1-NEXT: [[TMP87:%.*]] = getelementptr i8, ptr [[ARRAYIDX2]], i64 [[TMP11]] // CHECK1-NEXT: [[TMP81:%.*]] = getelementptr i8, ptr [[ARRAYIDX2]], i64 [[TMP11]]
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY22:%.*]] = icmp eq ptr [[ARRAYIDX2]], [[TMP87]] // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY22:%.*]] = icmp eq ptr [[ARRAYIDX2]], [[TMP81]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY22]], label [[OMP_ARRAYCPY_DONE35:%.*]], label [[OMP_ARRAYCPY_BODY23:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY22]], label [[OMP_ARRAYCPY_DONE35:%.*]], label [[OMP_ARRAYCPY_BODY23:%.*]]
// CHECK1: omp.arraycpy.body23: // CHECK1: omp.arraycpy.body23:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST24:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT33:%.*]], [[ATOMIC_EXIT:%.*]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST24:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT33:%.*]], [[ATOMIC_EXIT:%.*]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST25:%.*]] = phi ptr [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT32:%.*]], [[ATOMIC_EXIT]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST25:%.*]] = phi ptr [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT32:%.*]], [[ATOMIC_EXIT]] ]
// CHECK1-NEXT: [[TMP88:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST24]], align 1 // CHECK1-NEXT: [[TMP82:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST24]], align 1
// CHECK1-NEXT: [[CONV26:%.*]] = sext i8 [[TMP88]] to i32 // CHECK1-NEXT: [[CONV26:%.*]] = sext i8 [[TMP82]] to i32
// CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST25]] monotonic, align 1 // CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST25]] monotonic, align 1
// CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]] // CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]]
// CHECK1: atomic_cont: // CHECK1: atomic_cont:
// CHECK1-NEXT: [[TMP89:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY23]] ], [ [[TMP94:%.*]], [[ATOMIC_CONT]] ] // CHECK1-NEXT: [[TMP83:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY23]] ], [ [[TMP88:%.*]], [[ATOMIC_CONT]] ]
// CHECK1-NEXT: store i8 [[TMP89]], ptr [[_TMP27]], align 1 // CHECK1-NEXT: store i8 [[TMP83]], ptr [[_TMP27]], align 1
// CHECK1-NEXT: [[TMP90:%.*]] = load i8, ptr [[_TMP27]], align 1 // CHECK1-NEXT: [[TMP84:%.*]] = load i8, ptr [[_TMP27]], align 1
// CHECK1-NEXT: [[CONV28:%.*]] = sext i8 [[TMP90]] to i32 // CHECK1-NEXT: [[CONV28:%.*]] = sext i8 [[TMP84]] to i32
// CHECK1-NEXT: [[TMP91:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST24]], align 1 // CHECK1-NEXT: [[TMP85:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST24]], align 1
// CHECK1-NEXT: [[CONV29:%.*]] = sext i8 [[TMP91]] to i32 // CHECK1-NEXT: [[CONV29:%.*]] = sext i8 [[TMP85]] to i32
// CHECK1-NEXT: [[ADD30:%.*]] = add nsw i32 [[CONV28]], [[CONV29]] // CHECK1-NEXT: [[ADD30:%.*]] = add nsw i32 [[CONV28]], [[CONV29]]
// CHECK1-NEXT: [[CONV31:%.*]] = trunc i32 [[ADD30]] to i8 // CHECK1-NEXT: [[CONV31:%.*]] = trunc i32 [[ADD30]] to i8
// CHECK1-NEXT: store i8 [[CONV31]], ptr [[ATOMIC_TEMP]], align 1 // CHECK1-NEXT: store i8 [[CONV31]], ptr [[ATOMIC_TEMP]], align 1
// CHECK1-NEXT: [[TMP92:%.*]] = load i8, ptr [[ATOMIC_TEMP]], align 1 // CHECK1-NEXT: [[TMP86:%.*]] = load i8, ptr [[ATOMIC_TEMP]], align 1
// CHECK1-NEXT: [[TMP93:%.*]] = cmpxchg ptr [[OMP_ARRAYCPY_DESTELEMENTPAST25]], i8 [[TMP89]], i8 [[TMP92]] monotonic monotonic, align 1 // CHECK1-NEXT: [[TMP87:%.*]] = cmpxchg ptr [[OMP_ARRAYCPY_DESTELEMENTPAST25]], i8 [[TMP83]], i8 [[TMP86]] monotonic monotonic, align 1
// CHECK1-NEXT: [[TMP94]] = extractvalue { i8, i1 } [[TMP93]], 0 // CHECK1-NEXT: [[TMP88]] = extractvalue { i8, i1 } [[TMP87]], 0
// CHECK1-NEXT: [[TMP95:%.*]] = extractvalue { i8, i1 } [[TMP93]], 1 // CHECK1-NEXT: [[TMP89:%.*]] = extractvalue { i8, i1 } [[TMP87]], 1
// CHECK1-NEXT: br i1 [[TMP95]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]] // CHECK1-NEXT: br i1 [[TMP89]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]]
// CHECK1: atomic_exit: // CHECK1: atomic_exit:
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT32]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST25]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT32]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST25]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT33]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST24]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT33]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST24]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE34:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT32]], [[TMP87]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE34:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT32]], [[TMP81]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE34]], label [[OMP_ARRAYCPY_DONE35]], label [[OMP_ARRAYCPY_BODY23]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE34]], label [[OMP_ARRAYCPY_DONE35]], label [[OMP_ARRAYCPY_BODY23]]
// CHECK1: omp.arraycpy.done35: // CHECK1: omp.arraycpy.done35:
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: [[TMP96:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 // CHECK1-NEXT: [[TMP90:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK1-NEXT: call void @llvm.stackrestore(ptr [[TMP96]]) // CHECK1-NEXT: call void @llvm.stackrestore(ptr [[TMP90]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.red_init. // CHECK1-LABEL: define {{[^@]+}}@.red_init.
// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { // CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store i32 0, ptr [[TMP3]], align 4 // CHECK1-NEXT: store i32 0, ptr [[TMP2]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.red_comb. // CHECK1-LABEL: define {{[^@]+}}@.red_comb.
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP3]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP5]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP6]], [[TMP7]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP4]], [[TMP5]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP3]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP2]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.red_init..1 // CHECK1-LABEL: define {{[^@]+}}@.red_init..1
// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
▲ Show 20 Lines • Show All 128 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[TMP21:%.*]] = sub i64 [[TMP19]], [[TMP20]] // CHECK1-NEXT: [[TMP21:%.*]] = sub i64 [[TMP19]], [[TMP20]]
// CHECK1-NEXT: [[TMP22:%.*]] = sdiv exact i64 [[TMP21]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP22:%.*]] = sdiv exact i64 [[TMP21]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr i8, ptr [[VLA]], i64 [[TMP22]] // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr i8, ptr [[VLA]], i64 [[TMP22]]
// CHECK1-NEXT: store ptr [[_TMP6]], ptr [[_TMP5]], align 8 // CHECK1-NEXT: store ptr [[_TMP6]], ptr [[_TMP5]], align 8
// CHECK1-NEXT: store ptr [[TMP23]], ptr [[_TMP6]], align 8 // CHECK1-NEXT: store ptr [[TMP23]], ptr [[_TMP6]], align 8
// CHECK1-NEXT: [[DOTRD_INPUT_GEP_:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t.0], ptr [[DOTRD_INPUT_]], i64 0, i64 0 // CHECK1-NEXT: [[DOTRD_INPUT_GEP_:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t.0], ptr [[DOTRD_INPUT_]], i64 0, i64 0
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0:%.*]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 0 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0:%.*]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP24]], align 8 // CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP24]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 1 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP26]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[TMP25]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 2 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 2
// CHECK1-NEXT: store i64 4, ptr [[TMP28]], align 8 // CHECK1-NEXT: store i64 4, ptr [[TMP26]], align 8
// CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 3 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 3
// CHECK1-NEXT: store ptr @.red_init..4, ptr [[TMP29]], align 8 // CHECK1-NEXT: store ptr @.red_init..4, ptr [[TMP27]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 4 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 4
// CHECK1-NEXT: store ptr null, ptr [[TMP30]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP28]], align 8
// CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 5 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.red_comb..5, ptr [[TMP31]], align 8 // CHECK1-NEXT: store ptr @.red_comb..5, ptr [[TMP29]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 6 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_]], i32 0, i32 6
// CHECK1-NEXT: call void @llvm.memset.p0.i64(ptr align 8 [[TMP32]], i8 0, i64 4, i1 false) // CHECK1-NEXT: call void @llvm.memset.p0.i64(ptr align 8 [[TMP30]], i8 0, i64 4, i1 false)
// CHECK1-NEXT: [[DOTRD_INPUT_GEP_7:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t.0], ptr [[DOTRD_INPUT_]], i64 0, i64 1 // CHECK1-NEXT: [[DOTRD_INPUT_GEP_7:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t.0], ptr [[DOTRD_INPUT_]], i64 0, i64 1
// CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 0 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 0
// CHECK1-NEXT: [[TMP35:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8 // CHECK1-NEXT: [[TMP32:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8
// CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds ptr, ptr [[TMP35]], i64 0 // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds ptr, ptr [[TMP32]], i64 0
// CHECK1-NEXT: [[TMP36:%.*]] = load ptr, ptr [[ARRAYIDX8]], align 8 // CHECK1-NEXT: [[TMP33:%.*]] = load ptr, ptr [[ARRAYIDX8]], align 8
// CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds i8, ptr [[TMP36]], i64 0 // CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds i8, ptr [[TMP33]], i64 0
// CHECK1-NEXT: [[TMP37:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP34:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP38:%.*]] = sext i32 [[TMP37]] to i64 // CHECK1-NEXT: [[TMP35:%.*]] = sext i32 [[TMP34]] to i64
// CHECK1-NEXT: [[LB_ADD_LEN10:%.*]] = add nsw i64 -1, [[TMP38]] // CHECK1-NEXT: [[LB_ADD_LEN10:%.*]] = add nsw i64 -1, [[TMP35]]
// CHECK1-NEXT: [[TMP39:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8 // CHECK1-NEXT: [[TMP36:%.*]] = load ptr, ptr [[ARGV_ADDR]], align 8
// CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds ptr, ptr [[TMP39]], i64 9 // CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds ptr, ptr [[TMP36]], i64 9
// CHECK1-NEXT: [[TMP40:%.*]] = load ptr, ptr [[ARRAYIDX11]], align 8 // CHECK1-NEXT: [[TMP37:%.*]] = load ptr, ptr [[ARRAYIDX11]], align 8
// CHECK1-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds i8, ptr [[TMP40]], i64 [[LB_ADD_LEN10]] // CHECK1-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds i8, ptr [[TMP37]], i64 [[LB_ADD_LEN10]]
// CHECK1-NEXT: store ptr [[VLA]], ptr [[TMP34]], align 8 // CHECK1-NEXT: store ptr [[VLA]], ptr [[TMP31]], align 8
// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 1 // CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[ARRAYIDX9]], ptr [[TMP41]], align 8 // CHECK1-NEXT: store ptr [[ARRAYIDX9]], ptr [[TMP38]], align 8
// CHECK1-NEXT: [[TMP42:%.*]] = ptrtoint ptr [[ARRAYIDX12]] to i64 // CHECK1-NEXT: [[TMP39:%.*]] = ptrtoint ptr [[ARRAYIDX12]] to i64
// CHECK1-NEXT: [[TMP43:%.*]] = ptrtoint ptr [[ARRAYIDX9]] to i64 // CHECK1-NEXT: [[TMP40:%.*]] = ptrtoint ptr [[ARRAYIDX9]] to i64
// CHECK1-NEXT: [[TMP44:%.*]] = sub i64 [[TMP42]], [[TMP43]] // CHECK1-NEXT: [[TMP41:%.*]] = sub i64 [[TMP39]], [[TMP40]]
// CHECK1-NEXT: [[TMP45:%.*]] = sdiv exact i64 [[TMP44]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP42:%.*]] = sdiv exact i64 [[TMP41]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP46:%.*]] = add nuw i64 [[TMP45]], 1 // CHECK1-NEXT: [[TMP43:%.*]] = add nuw i64 [[TMP42]], 1
// CHECK1-NEXT: [[TMP47:%.*]] = mul nuw i64 [[TMP46]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP44:%.*]] = mul nuw i64 [[TMP43]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 2 // CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 2
// CHECK1-NEXT: store i64 [[TMP47]], ptr [[TMP48]], align 8 // CHECK1-NEXT: store i64 [[TMP44]], ptr [[TMP45]], align 8
// CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 3 // CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 3
// CHECK1-NEXT: store ptr @.red_init..6, ptr [[TMP49]], align 8 // CHECK1-NEXT: store ptr @.red_init..6, ptr [[TMP46]], align 8
// CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 4 // CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 4
// CHECK1-NEXT: store ptr null, ptr [[TMP50]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP47]], align 8
// CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 5 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.red_comb..7, ptr [[TMP51]], align 8 // CHECK1-NEXT: store ptr @.red_comb..7, ptr [[TMP48]], align 8
// CHECK1-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 6 // CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T_0]], ptr [[DOTRD_INPUT_GEP_7]], i32 0, i32 6
// CHECK1-NEXT: store i32 1, ptr [[TMP52]], align 8 // CHECK1-NEXT: store i32 1, ptr [[TMP49]], align 8
// CHECK1-NEXT: [[TMP50:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP51:%.*]] = load i32, ptr [[TMP50]], align 4
// CHECK1-NEXT: [[TMP52:%.*]] = call ptr @__kmpc_taskred_modifier_init(ptr @[[GLOB1]], i32 [[TMP51]], i32 1, i32 2, ptr [[DOTRD_INPUT_]])
// CHECK1-NEXT: store ptr [[TMP52]], ptr [[DOTTASK_RED_]], align 8
// CHECK1-NEXT: [[TMP53:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP53:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP54:%.*]] = load i32, ptr [[TMP53]], align 4 // CHECK1-NEXT: [[TMP54:%.*]] = load i32, ptr [[TMP53]], align 4
// CHECK1-NEXT: [[TMP56:%.*]] = call ptr @__kmpc_taskred_modifier_init(ptr @[[GLOB1]], i32 [[TMP54]], i32 1, i32 2, ptr [[DOTRD_INPUT_]]) // CHECK1-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB3:[0-9]+]], i32 [[TMP54]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK1-NEXT: store ptr [[TMP56]], ptr [[DOTTASK_RED_]], align 8 // CHECK1-NEXT: [[TMP55:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK1-NEXT: [[TMP57:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i64 [[TMP55]], 9
// CHECK1-NEXT: [[TMP58:%.*]] = load i32, ptr [[TMP57]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB3:[0-9]+]], i32 [[TMP58]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK1-NEXT: [[TMP59:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i64 [[TMP59]], 9
// CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1: cond.true: // CHECK1: cond.true:
// CHECK1-NEXT: br label [[COND_END:%.*]] // CHECK1-NEXT: br label [[COND_END:%.*]]
// CHECK1: cond.false: // CHECK1: cond.false:
// CHECK1-NEXT: [[TMP60:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8 // CHECK1-NEXT: [[TMP56:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK1-NEXT: br label [[COND_END]] // CHECK1-NEXT: br label [[COND_END]]
// CHECK1: cond.end: // CHECK1: cond.end:
// CHECK1-NEXT: [[COND:%.*]] = phi i64 [ 9, [[COND_TRUE]] ], [ [[TMP60]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i64 [ 9, [[COND_TRUE]] ], [ [[TMP56]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i64 [[COND]], ptr [[DOTOMP_UB]], align 8 // CHECK1-NEXT: store i64 [[COND]], ptr [[DOTOMP_UB]], align 8
// CHECK1-NEXT: [[TMP61:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8 // CHECK1-NEXT: [[TMP57:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK1-NEXT: store i64 [[TMP61]], ptr [[DOTOMP_IV]], align 8 // CHECK1-NEXT: store i64 [[TMP57]], ptr [[DOTOMP_IV]], align 8
// 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: [[TMP62:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK1-NEXT: [[TMP58:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK1-NEXT: [[TMP63:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8 // CHECK1-NEXT: [[TMP59:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK1-NEXT: [[CMP13:%.*]] = icmp sle i64 [[TMP62]], [[TMP63]] // CHECK1-NEXT: [[CMP13:%.*]] = icmp sle i64 [[TMP58]], [[TMP59]]
// CHECK1-NEXT: br i1 [[CMP13]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]] // CHECK1-NEXT: br i1 [[CMP13]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1: omp.inner.for.cond.cleanup: // CHECK1: omp.inner.for.cond.cleanup:
// CHECK1-NEXT: br label [[OMP_INNER_FOR_END:%.*]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body: // CHECK1: omp.inner.for.body:
// CHECK1-NEXT: [[TMP64:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK1-NEXT: [[TMP60:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i64 [[TMP64]], 1 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i64 [[TMP60]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL]]
// CHECK1-NEXT: store i64 [[ADD]], ptr [[I]], align 8 // CHECK1-NEXT: store i64 [[ADD]], ptr [[I]], align 8
// CHECK1-NEXT: [[TMP65:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 0 // CHECK1-NEXT: [[TMP61:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[DOTTASK_RED_]], ptr [[TMP65]], align 8 // CHECK1-NEXT: store ptr [[DOTTASK_RED_]], ptr [[TMP61]], align 8
// CHECK1-NEXT: [[TMP66:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 1 // CHECK1-NEXT: [[TMP62:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP66]], align 8 // CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP62]], align 8
// CHECK1-NEXT: [[TMP67:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 2 // CHECK1-NEXT: [[TMP63:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[AGG_CAPTURED]], i32 0, i32 2
// CHECK1-NEXT: [[TMP68:%.*]] = load ptr, ptr [[_TMP5]], align 8 // CHECK1-NEXT: [[TMP64:%.*]] = load ptr, ptr [[_TMP5]], align 8
// CHECK1-NEXT: store ptr [[TMP68]], ptr [[TMP67]], align 8 // CHECK1-NEXT: store ptr [[TMP64]], ptr [[TMP63]], align 8
// CHECK1-NEXT: [[TMP69:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP65:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP70:%.*]] = load i32, ptr [[TMP69]], align 4 // CHECK1-NEXT: [[TMP66:%.*]] = load i32, ptr [[TMP65]], align 4
// CHECK1-NEXT: [[TMP71:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP70]], i32 1, i64 48, i64 24, ptr @.omp_task_entry.) // CHECK1-NEXT: [[TMP67:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB1]], i32 [[TMP66]], i32 1, i64 48, i64 24, ptr @.omp_task_entry.)
// CHECK1-NEXT: [[TMP73:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP71]], i32 0, i32 0 // CHECK1-NEXT: [[TMP68:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP67]], i32 0, i32 0
// CHECK1-NEXT: [[TMP74:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP73]], i32 0, i32 0 // CHECK1-NEXT: [[TMP69:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP68]], i32 0, i32 0
// CHECK1-NEXT: [[TMP75:%.*]] = load ptr, ptr [[TMP74]], align 8 // CHECK1-NEXT: [[TMP70:%.*]] = load ptr, ptr [[TMP69]], align 8
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP75]], ptr align 8 [[AGG_CAPTURED]], i64 24, i1 false) // CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[TMP70]], ptr align 8 [[AGG_CAPTURED]], i64 24, i1 false)
// CHECK1-NEXT: [[TMP77:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP71]], i32 0, i32 1 // CHECK1-NEXT: [[TMP71:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP67]], i32 0, i32 1
// CHECK1-NEXT: [[TMP78:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP77]], i32 0, i32 0 // CHECK1-NEXT: [[TMP72:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], ptr [[TMP71]], i32 0, i32 0
// CHECK1-NEXT: [[TMP79:%.*]] = load ptr, ptr [[DOTTASK_RED_]], align 8 // CHECK1-NEXT: [[TMP73:%.*]] = load ptr, ptr [[DOTTASK_RED_]], align 8
// CHECK1-NEXT: store ptr [[TMP79]], ptr [[TMP78]], align 8 // CHECK1-NEXT: store ptr [[TMP73]], ptr [[TMP72]], align 8
// CHECK1-NEXT: [[TMP80:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP74:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP81:%.*]] = load i32, ptr [[TMP80]], align 4 // CHECK1-NEXT: [[TMP75:%.*]] = load i32, ptr [[TMP74]], align 4
// CHECK1-NEXT: [[TMP82:%.*]] = call i32 @__kmpc_omp_task(ptr @[[GLOB1]], i32 [[TMP81]], ptr [[TMP71]]) // CHECK1-NEXT: [[TMP76:%.*]] = call i32 @__kmpc_omp_task(ptr @[[GLOB1]], i32 [[TMP75]], ptr [[TMP67]])
// 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: [[TMP83:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8 // CHECK1-NEXT: [[TMP77:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8
// CHECK1-NEXT: [[ADD14:%.*]] = add nsw i64 [[TMP83]], 1 // CHECK1-NEXT: [[ADD14:%.*]] = add nsw i64 [[TMP77]], 1
// CHECK1-NEXT: store i64 [[ADD14]], ptr [[DOTOMP_IV]], align 8 // CHECK1-NEXT: store i64 [[ADD14]], ptr [[DOTOMP_IV]], align 8
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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: [[TMP84:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP78:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP85:%.*]] = load i32, ptr [[TMP84]], align 4 // CHECK1-NEXT: [[TMP79:%.*]] = load i32, ptr [[TMP78]], align 4
// CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP85]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP79]])
// CHECK1-NEXT: [[TMP80:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP81:%.*]] = load i32, ptr [[TMP80]], align 4
// CHECK1-NEXT: call void @__kmpc_task_reduction_modifier_fini(ptr @[[GLOB1]], i32 [[TMP81]], i32 1)
// CHECK1-NEXT: [[TMP82:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP82]], align 8
// CHECK1-NEXT: [[TMP83:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT: store ptr [[VLA]], ptr [[TMP83]], align 8
// CHECK1-NEXT: [[TMP84:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK1-NEXT: [[TMP85:%.*]] = inttoptr i64 [[TMP13]] to ptr
// CHECK1-NEXT: store ptr [[TMP85]], ptr [[TMP84]], align 8
// CHECK1-NEXT: [[TMP86:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP86:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP87:%.*]] = load i32, ptr [[TMP86]], align 4 // CHECK1-NEXT: [[TMP87:%.*]] = load i32, ptr [[TMP86]], align 4
// CHECK1-NEXT: call void @__kmpc_task_reduction_modifier_fini(ptr @[[GLOB1]], i32 [[TMP87]], i32 1) // CHECK1-NEXT: [[TMP88:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB4]], i32 [[TMP87]], i32 2, i64 24, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: [[TMP88:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: switch i32 [[TMP88]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: store ptr [[ARGC1]], ptr [[TMP88]], align 8
// CHECK1-NEXT: [[TMP90:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT: store ptr [[VLA]], ptr [[TMP90]], align 8
// CHECK1-NEXT: [[TMP91:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK1-NEXT: [[TMP92:%.*]] = inttoptr i64 [[TMP13]] to ptr
// CHECK1-NEXT: store ptr [[TMP92]], ptr [[TMP91]], align 8
// CHECK1-NEXT: [[TMP93:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP94:%.*]] = load i32, ptr [[TMP93]], align 4
// CHECK1-NEXT: [[TMP96:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB4]], i32 [[TMP94]], i32 2, i64 24, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP96]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP97:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP89:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP98:%.*]] = load i32, ptr [[ARGC1]], align 4 // CHECK1-NEXT: [[TMP90:%.*]] = load i32, ptr [[ARGC1]], align 4
// CHECK1-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP97]], [[TMP98]] // CHECK1-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP89]], [[TMP90]]
// CHECK1-NEXT: store i32 [[ADD15]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD15]], ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP99:%.*]] = getelementptr i8, ptr [[ARRAYIDX2]], i64 [[TMP13]] // CHECK1-NEXT: [[TMP91:%.*]] = getelementptr i8, ptr [[ARRAYIDX2]], i64 [[TMP13]]
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX2]], [[TMP99]] // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX2]], [[TMP91]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE22:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE22:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1: omp.arraycpy.body: // CHECK1: omp.arraycpy.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST16:%.*]] = phi ptr [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT20:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST16:%.*]] = phi ptr [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT20:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[TMP100:%.*]] = load i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1 // CHECK1-NEXT: [[TMP92:%.*]] = load i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1
// CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP100]] to i32 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP92]] to i32
// CHECK1-NEXT: [[TMP101:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 // CHECK1-NEXT: [[TMP93:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
// CHECK1-NEXT: [[CONV17:%.*]] = sext i8 [[TMP101]] to i32 // CHECK1-NEXT: [[CONV17:%.*]] = sext i8 [[TMP93]] to i32
// CHECK1-NEXT: [[ADD18:%.*]] = add nsw i32 [[CONV]], [[CONV17]] // CHECK1-NEXT: [[ADD18:%.*]] = add nsw i32 [[CONV]], [[CONV17]]
// CHECK1-NEXT: [[CONV19:%.*]] = trunc i32 [[ADD18]] to i8 // CHECK1-NEXT: [[CONV19:%.*]] = trunc i32 [[ADD18]] to i8
// CHECK1-NEXT: store i8 [[CONV19]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1 // CHECK1-NEXT: store i8 [[CONV19]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT20]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST16]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT20]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST16]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE21:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT20]], [[TMP99]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE21:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT20]], [[TMP91]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_DONE22]], label [[OMP_ARRAYCPY_BODY]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_DONE22]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1: omp.arraycpy.done22: // CHECK1: omp.arraycpy.done22:
// CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB4]], i32 [[TMP94]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB4]], i32 [[TMP87]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP102:%.*]] = load i32, ptr [[ARGC1]], align 4 // CHECK1-NEXT: [[TMP94:%.*]] = load i32, ptr [[ARGC1]], align 4
// CHECK1-NEXT: [[TMP103:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP102]] monotonic, align 4 // CHECK1-NEXT: [[TMP95:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP94]] monotonic, align 4
// CHECK1-NEXT: [[TMP104:%.*]] = getelementptr i8, ptr [[ARRAYIDX2]], i64 [[TMP13]] // CHECK1-NEXT: [[TMP96:%.*]] = getelementptr i8, ptr [[ARRAYIDX2]], i64 [[TMP13]]
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY23:%.*]] = icmp eq ptr [[ARRAYIDX2]], [[TMP104]] // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY23:%.*]] = icmp eq ptr [[ARRAYIDX2]], [[TMP96]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY23]], label [[OMP_ARRAYCPY_DONE36:%.*]], label [[OMP_ARRAYCPY_BODY24:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY23]], label [[OMP_ARRAYCPY_DONE36:%.*]], label [[OMP_ARRAYCPY_BODY24:%.*]]
// CHECK1: omp.arraycpy.body24: // CHECK1: omp.arraycpy.body24:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST25:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT34:%.*]], [[ATOMIC_EXIT:%.*]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST25:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT34:%.*]], [[ATOMIC_EXIT:%.*]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST26:%.*]] = phi ptr [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT33:%.*]], [[ATOMIC_EXIT]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST26:%.*]] = phi ptr [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT33:%.*]], [[ATOMIC_EXIT]] ]
// CHECK1-NEXT: [[TMP105:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1 // CHECK1-NEXT: [[TMP97:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1
// CHECK1-NEXT: [[CONV27:%.*]] = sext i8 [[TMP105]] to i32 // CHECK1-NEXT: [[CONV27:%.*]] = sext i8 [[TMP97]] to i32
// CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST26]] monotonic, align 1 // CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST26]] monotonic, align 1
// CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]] // CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]]
// CHECK1: atomic_cont: // CHECK1: atomic_cont:
// CHECK1-NEXT: [[TMP106:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY24]] ], [ [[TMP111:%.*]], [[ATOMIC_CONT]] ] // CHECK1-NEXT: [[TMP98:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY24]] ], [ [[TMP103:%.*]], [[ATOMIC_CONT]] ]
// CHECK1-NEXT: store i8 [[TMP106]], ptr [[_TMP28]], align 1 // CHECK1-NEXT: store i8 [[TMP98]], ptr [[_TMP28]], align 1
// CHECK1-NEXT: [[TMP107:%.*]] = load i8, ptr [[_TMP28]], align 1 // CHECK1-NEXT: [[TMP99:%.*]] = load i8, ptr [[_TMP28]], align 1
// CHECK1-NEXT: [[CONV29:%.*]] = sext i8 [[TMP107]] to i32 // CHECK1-NEXT: [[CONV29:%.*]] = sext i8 [[TMP99]] to i32
// CHECK1-NEXT: [[TMP108:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1 // CHECK1-NEXT: [[TMP100:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1
// CHECK1-NEXT: [[CONV30:%.*]] = sext i8 [[TMP108]] to i32 // CHECK1-NEXT: [[CONV30:%.*]] = sext i8 [[TMP100]] to i32
// CHECK1-NEXT: [[ADD31:%.*]] = add nsw i32 [[CONV29]], [[CONV30]] // CHECK1-NEXT: [[ADD31:%.*]] = add nsw i32 [[CONV29]], [[CONV30]]
// CHECK1-NEXT: [[CONV32:%.*]] = trunc i32 [[ADD31]] to i8 // CHECK1-NEXT: [[CONV32:%.*]] = trunc i32 [[ADD31]] to i8
// CHECK1-NEXT: store i8 [[CONV32]], ptr [[ATOMIC_TEMP]], align 1 // CHECK1-NEXT: store i8 [[CONV32]], ptr [[ATOMIC_TEMP]], align 1
// CHECK1-NEXT: [[TMP109:%.*]] = load i8, ptr [[ATOMIC_TEMP]], align 1 // CHECK1-NEXT: [[TMP101:%.*]] = load i8, ptr [[ATOMIC_TEMP]], align 1
// CHECK1-NEXT: [[TMP110:%.*]] = cmpxchg ptr [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i8 [[TMP106]], i8 [[TMP109]] monotonic monotonic, align 1 // CHECK1-NEXT: [[TMP102:%.*]] = cmpxchg ptr [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i8 [[TMP98]], i8 [[TMP101]] monotonic monotonic, align 1
// CHECK1-NEXT: [[TMP111]] = extractvalue { i8, i1 } [[TMP110]], 0 // CHECK1-NEXT: [[TMP103]] = extractvalue { i8, i1 } [[TMP102]], 0
// CHECK1-NEXT: [[TMP112:%.*]] = extractvalue { i8, i1 } [[TMP110]], 1 // CHECK1-NEXT: [[TMP104:%.*]] = extractvalue { i8, i1 } [[TMP102]], 1
// CHECK1-NEXT: br i1 [[TMP112]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]] // CHECK1-NEXT: br i1 [[TMP104]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]]
// CHECK1: atomic_exit: // CHECK1: atomic_exit:
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT33]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT33]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT34]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST25]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT34]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST25]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE35:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT33]], [[TMP104]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE35:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT33]], [[TMP96]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE35]], label [[OMP_ARRAYCPY_DONE36]], label [[OMP_ARRAYCPY_BODY24]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE35]], label [[OMP_ARRAYCPY_DONE36]], label [[OMP_ARRAYCPY_BODY24]]
// CHECK1: omp.arraycpy.done36: // CHECK1: omp.arraycpy.done36:
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: [[TMP113:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 // CHECK1-NEXT: [[TMP105:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK1-NEXT: call void @llvm.stackrestore(ptr [[TMP113]]) // CHECK1-NEXT: call void @llvm.stackrestore(ptr [[TMP105]])
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.red_init..4 // CHECK1-LABEL: define {{[^@]+}}@.red_init..4
// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store i32 0, ptr [[TMP3]], align 4 // CHECK1-NEXT: store i32 0, ptr [[TMP2]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.red_comb..5 // CHECK1-LABEL: define {{[^@]+}}@.red_comb..5
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP3]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP5]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP6]], [[TMP7]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP4]], [[TMP5]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP3]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP2]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.red_init..6 // CHECK1-LABEL: define {{[^@]+}}@.red_init..6
// CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noalias noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
▲ Show 20 Lines • Show All 78 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK1-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4
// CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP3]], i32 0, i32 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP3]], i32 0, i32 0
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP3]], i32 0, i32 1 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], ptr [[TMP3]], i32 0, i32 1
// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META3:![0-9]+]]) // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META3:![0-9]+]])
// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META6:![0-9]+]]) // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META6:![0-9]+]])
// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META8:![0-9]+]]) // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META8:![0-9]+]])
// CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]]) // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]])
// CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: store ptr [[TMP9]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP8]], ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: store ptr @.omp_task_privates_map., ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr @.omp_task_privates_map., ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: call void [[TMP13]](ptr [[TMP14]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR6]] // CHECK1-NEXT: call void [[TMP10]](ptr [[TMP11]], ptr [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR6]]
// CHECK1-NEXT: [[TMP16:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias !12 // CHECK1-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias !12
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], ptr [[TMP12]], i32 0, i32 1 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], ptr [[TMP9]], i32 0, i32 1
// CHECK1-NEXT: [[TMP18:%.*]] = load ptr, ptr [[TMP17]], align 8 // CHECK1-NEXT: [[TMP14:%.*]] = load ptr, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = load ptr, ptr [[TMP16]], align 8 // CHECK1-NEXT: [[TMP15:%.*]] = load ptr, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12
// CHECK1-NEXT: [[TMP22:%.*]] = call ptr @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], ptr [[TMP19]], ptr [[TMP18]]) // CHECK1-NEXT: [[TMP17:%.*]] = call ptr @__kmpc_task_reduction_get_th_data(i32 [[TMP16]], ptr [[TMP15]], ptr [[TMP14]])
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP12]], i32 0, i32 2 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP9]], i32 0, i32 2
// CHECK1-NEXT: [[TMP24:%.*]] = load ptr, ptr [[TMP23]], align 8 // CHECK1-NEXT: [[TMP19:%.*]] = load ptr, ptr [[TMP18]], align 8
// CHECK1-NEXT: [[TMP25:%.*]] = load ptr, ptr [[TMP24]], align 8 // CHECK1-NEXT: [[TMP20:%.*]] = load ptr, ptr [[TMP19]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP12]], i32 0, i32 1 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP9]], i32 0, i32 1
// CHECK1-NEXT: [[TMP27:%.*]] = load ptr, ptr [[TMP26]], align 8 // CHECK1-NEXT: [[TMP22:%.*]] = load ptr, ptr [[TMP21]], align 8
// CHECK1-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4 // CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
// CHECK1-NEXT: [[TMP29:%.*]] = sext i32 [[TMP28]] to i64 // CHECK1-NEXT: [[TMP24:%.*]] = sext i32 [[TMP23]] to i64
// CHECK1-NEXT: [[LB_ADD_LEN_I:%.*]] = add nsw i64 -1, [[TMP29]] // CHECK1-NEXT: [[LB_ADD_LEN_I:%.*]] = add nsw i64 -1, [[TMP24]]
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP12]], i32 0, i32 2 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP9]], i32 0, i32 2
// CHECK1-NEXT: [[TMP31:%.*]] = load ptr, ptr [[TMP30]], align 8 // CHECK1-NEXT: [[TMP26:%.*]] = load ptr, ptr [[TMP25]], align 8
// CHECK1-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds ptr, ptr [[TMP31]], i64 9 // CHECK1-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds ptr, ptr [[TMP26]], i64 9
// CHECK1-NEXT: [[TMP32:%.*]] = load ptr, ptr [[ARRAYIDX2_I]], align 8 // CHECK1-NEXT: [[TMP27:%.*]] = load ptr, ptr [[ARRAYIDX2_I]], align 8
// CHECK1-NEXT: [[ARRAYIDX3_I:%.*]] = getelementptr inbounds i8, ptr [[TMP32]], i64 [[LB_ADD_LEN_I]] // CHECK1-NEXT: [[ARRAYIDX3_I:%.*]] = getelementptr inbounds i8, ptr [[TMP27]], i64 [[LB_ADD_LEN_I]]
// CHECK1-NEXT: [[TMP33:%.*]] = ptrtoint ptr [[ARRAYIDX3_I]] to i64 // CHECK1-NEXT: [[TMP28:%.*]] = ptrtoint ptr [[ARRAYIDX3_I]] to i64
// CHECK1-NEXT: [[TMP34:%.*]] = ptrtoint ptr [[TMP25]] to i64 // CHECK1-NEXT: [[TMP29:%.*]] = ptrtoint ptr [[TMP20]] to i64
// CHECK1-NEXT: [[TMP35:%.*]] = sub i64 [[TMP33]], [[TMP34]] // CHECK1-NEXT: [[TMP30:%.*]] = sub i64 [[TMP28]], [[TMP29]]
// CHECK1-NEXT: [[TMP36:%.*]] = sdiv exact i64 [[TMP35]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP31:%.*]] = sdiv exact i64 [[TMP30]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP37:%.*]] = add nuw i64 [[TMP36]], 1 // CHECK1-NEXT: [[TMP32:%.*]] = add nuw i64 [[TMP31]], 1
// CHECK1-NEXT: [[TMP38:%.*]] = mul nuw i64 [[TMP37]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP33:%.*]] = mul nuw i64 [[TMP32]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: store i64 [[TMP37]], ptr @{{reduction_size[.].+[.]}}, align 8, !noalias !12 // CHECK1-NEXT: store i64 [[TMP32]], ptr @{{reduction_size[.].+[.]}}, align 8, !noalias !12
// CHECK1-NEXT: [[TMP39:%.*]] = load ptr, ptr [[TMP16]], align 8 // CHECK1-NEXT: [[TMP34:%.*]] = load ptr, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = call ptr @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], ptr [[TMP39]], ptr [[TMP25]]) // CHECK1-NEXT: [[TMP35:%.*]] = call ptr @__kmpc_task_reduction_get_th_data(i32 [[TMP16]], ptr [[TMP34]], ptr [[TMP20]])
// CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP12]], i32 0, i32 2 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT_ANON]], ptr [[TMP9]], i32 0, i32 2
// CHECK1-NEXT: [[TMP42:%.*]] = load ptr, ptr [[TMP41]], align 8 // CHECK1-NEXT: [[TMP37:%.*]] = load ptr, ptr [[TMP36]], align 8
// CHECK1-NEXT: [[TMP43:%.*]] = load ptr, ptr [[TMP42]], align 8 // CHECK1-NEXT: [[TMP38:%.*]] = load ptr, ptr [[TMP37]], align 8
// CHECK1-NEXT: [[TMP44:%.*]] = ptrtoint ptr [[TMP43]] to i64 // CHECK1-NEXT: [[TMP39:%.*]] = ptrtoint ptr [[TMP38]] to i64
// CHECK1-NEXT: [[TMP45:%.*]] = ptrtoint ptr [[TMP25]] to i64 // CHECK1-NEXT: [[TMP40:%.*]] = ptrtoint ptr [[TMP20]] to i64
// CHECK1-NEXT: [[TMP46:%.*]] = sub i64 [[TMP44]], [[TMP45]] // CHECK1-NEXT: [[TMP41:%.*]] = sub i64 [[TMP39]], [[TMP40]]
// CHECK1-NEXT: [[TMP47:%.*]] = sdiv exact i64 [[TMP46]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) // CHECK1-NEXT: [[TMP42:%.*]] = sdiv exact i64 [[TMP41]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64)
// CHECK1-NEXT: [[TMP48:%.*]] = getelementptr i8, ptr [[TMP40]], i64 [[TMP47]] // CHECK1-NEXT: [[TMP43:%.*]] = getelementptr i8, ptr [[TMP35]], i64 [[TMP42]]
// CHECK1-NEXT: store ptr [[TMP4_I]], ptr [[TMP_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP4_I]], ptr [[TMP_I]], align 8, !noalias !12
// CHECK1-NEXT: store ptr [[TMP48]], ptr [[TMP4_I]], align 8, !noalias !12 // CHECK1-NEXT: store ptr [[TMP43]], ptr [[TMP4_I]], align 8, !noalias !12
// CHECK1-NEXT: ret i32 0 // CHECK1-NEXT: ret i32 0
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP3]], i64 0, i64 1
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP4]], i64 0, i64 1 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 1
// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 2
// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[TMP12]], align 8 // CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 1 // CHECK1-NEXT: [[TMP14:%.*]] = ptrtoint ptr [[TMP13]] to i64
// CHECK1-NEXT: [[TMP15:%.*]] = load ptr, ptr [[TMP14]], align 8 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 2 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP17:%.*]] = load ptr, ptr [[TMP16]], align 8 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP15]], [[TMP16]]
// CHECK1-NEXT: [[TMP18:%.*]] = ptrtoint ptr [[TMP17]] to i64 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[TMP11]], i64 [[TMP14]]
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP11]], [[TMP17]]
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[TMP15]], i64 [[TMP18]]
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP15]], [[TMP21]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1: omp.arraycpy.body: // CHECK1: omp.arraycpy.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP13]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP9]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP15]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[TMP22:%.*]] = load i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 // CHECK1-NEXT: [[TMP18:%.*]] = load i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP22]] to i32 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP18]] to i32
// CHECK1-NEXT: [[TMP23:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 // CHECK1-NEXT: [[TMP19:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
// CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP23]] to i32 // CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP19]] to i32
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]] // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]]
// CHECK1-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8 // CHECK1-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8
// CHECK1-NEXT: store i8 [[CONV4]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 // CHECK1-NEXT: store i8 [[CONV4]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP21]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP17]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1: omp.arraycpy.done5: // CHECK1: omp.arraycpy.done5:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.9 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.9
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP3]], i64 0, i64 1
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP4]], i64 0, i64 1 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 1
// CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 2
// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[TMP12]], align 8 // CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 1 // CHECK1-NEXT: [[TMP14:%.*]] = ptrtoint ptr [[TMP13]] to i64
// CHECK1-NEXT: [[TMP15:%.*]] = load ptr, ptr [[TMP14]], align 8 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 2 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP17:%.*]] = load ptr, ptr [[TMP16]], align 8 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP15]], [[TMP16]]
// CHECK1-NEXT: [[TMP18:%.*]] = ptrtoint ptr [[TMP17]] to i64 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr i8, ptr [[TMP11]], i64 [[TMP14]]
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP11]], [[TMP17]]
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, ptr [[TMP15]], i64 [[TMP18]]
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP15]], [[TMP21]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1: omp.arraycpy.body: // CHECK1: omp.arraycpy.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP13]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP9]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP15]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP11]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[TMP22:%.*]] = load i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 // CHECK1-NEXT: [[TMP18:%.*]] = load i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP22]] to i32 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP18]] to i32
// CHECK1-NEXT: [[TMP23:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 // CHECK1-NEXT: [[TMP19:%.*]] = load i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1
// CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP23]] to i32 // CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP19]] to i32
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]] // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]]
// CHECK1-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8 // CHECK1-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8
// CHECK1-NEXT: store i8 [[CONV4]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 // CHECK1-NEXT: store i8 [[CONV4]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP21]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP17]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1: omp.arraycpy.done5: // CHECK1: omp.arraycpy.done5:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //

clang/test/OpenMP/target_teams_distribute_parallel_for_simd_collapse_codegen.cpp

Show First 20 Lines • Show All 115 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 8 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[A]], ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr [[A]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 56088, ptr [[TMP15]], align 8 // CHECK1-NEXT: store i64 56088, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP15]], 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]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 // CHECK1-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i64 0, i64 0 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i64 0, i64 0
// CHECK1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 0 // CHECK1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 0
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4
// CHECK1-NEXT: ret i32 [[TMP18]] // CHECK1-NEXT: ret i32 [[TMP16]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28
// CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
Show All 37 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]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4:![0-9]+]]
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP4]]
// 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, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @.omp_outlined..1, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]]), !llvm.access.group !4 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @.omp_outlined..1, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]]), !llvm.access.group [[ACC_GRP4]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[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:%.*]]
▲ Show 20 Lines • Show All 51 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]], ptr [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP7]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !8 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8:![0-9]+]]
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !8 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP8]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !8 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8]]
// 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]], ptr [[I]], align 4, !llvm.access.group !8 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP8]]
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !8 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8]]
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !8 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8]]
// 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]], ptr [[J]], align 4, !llvm.access.group !8 // CHECK1-NEXT: store i32 [[ADD7]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP8]]
// CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !8 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP8]]
// 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]], ptr [[A]], i64 0, i64 [[IDXPROM]] // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group !8 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP8]]
// CHECK1-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP14]] to i64 // CHECK1-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP14]] to i64
// CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM8]] // CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM8]]
// CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX9]], align 4, !llvm.access.group !8 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX9]], align 4, !llvm.access.group [[ACC_GRP8]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !8 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8]]
// CHECK1-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP15]], 1 // CHECK1-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK1-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !8 // CHECK1-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP9:![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(ptr @[[GLOB1]], i32 [[TMP4]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP4]])
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[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:%.*]]
Show All 29 Lines
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[A]], ptr [[TMP2]], align 4 // CHECK3-NEXT: store ptr [[A]], ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 56088, ptr [[TMP15]], align 8 // CHECK3-NEXT: store i64 56088, ptr [[TMP13]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP15]], 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]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 // CHECK3-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i32 0, i32 0 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i32 0, i32 0
// CHECK3-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 0 // CHECK3-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 0
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4
// CHECK3-NEXT: ret i32 [[TMP18]] // CHECK3-NEXT: ret i32 [[TMP16]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28
// CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
Show All 37 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]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !5 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5:![0-9]+]]
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !5 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP5]]
// 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, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !5 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !5 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @.omp_outlined..1, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]]), !llvm.access.group !5 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @.omp_outlined..1, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]]), !llvm.access.group [[ACC_GRP5]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !5 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group !5 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !5 // CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[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:%.*]]
▲ Show 20 Lines • Show All 49 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]], ptr [[DOTOMP_UB]], align 4 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[TMP7]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !9 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9:![0-9]+]]
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !9 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP9]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !9 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9]]
// 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]], ptr [[I]], align 4, !llvm.access.group !9 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !9 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !9 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9]]
// 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]], ptr [[J]], align 4, !llvm.access.group !9 // CHECK3-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !9 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i32 0, i32 [[TMP13]] // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i32 0, i32 [[TMP13]]
// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group !9 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK3-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP14]] // CHECK3-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP14]]
// CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group !9 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group [[ACC_GRP9]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !9 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP15]], 1 // CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK3-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !9 // CHECK3-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![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(ptr @[[GLOB1]], i32 [[TMP4]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP4]])
// CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[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:%.*]]
Show All 34 Lines
// CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK5-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK5-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 56087, ptr [[DOTOMP_UB]], align 4 // CHECK5-NEXT: store i32 56087, ptr [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond: // CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2:![0-9]+]]
// CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] // CHECK5-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK5-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body: // CHECK5: omp.inner.for.body:
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 456 // CHECK5-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 456
// CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !2 // CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP5]], 456 // CHECK5-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP5]], 456
// CHECK5-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456 // CHECK5-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456
// CHECK5-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL4]] // CHECK5-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL4]]
// CHECK5-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK5-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK5-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]] // CHECK5-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK5-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group !2 // CHECK5-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 // CHECK5-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP6]] to i64 // CHECK5-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP6]] to i64
// CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i64 0, i64 [[IDXPROM]] // CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i64 0, i64 [[IDXPROM]]
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP7]] to i64 // CHECK5-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP7]] to i64
// CHECK5-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]] // CHECK5-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]]
// CHECK5-NEXT: store i32 0, ptr [[ARRAYIDX8]], align 4, !llvm.access.group !2 // CHECK5-NEXT: store i32 0, ptr [[ARRAYIDX8]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK5: omp.body.continue: // CHECK5: omp.body.continue:
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc: // CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP8]], 1 // CHECK5-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK5-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK5-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK5: omp.inner.for.end: // CHECK5: omp.inner.for.end:
// CHECK5-NEXT: store i32 123, ptr [[I]], align 4 // CHECK5-NEXT: store i32 123, ptr [[I]], align 4
// CHECK5-NEXT: store i32 456, ptr [[J]], align 4 // CHECK5-NEXT: store i32 456, ptr [[J]], align 4
// CHECK5-NEXT: [[A10:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 // CHECK5-NEXT: [[A10:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK5-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A10]], i64 0, i64 0 // CHECK5-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A10]], i64 0, i64 0
// CHECK5-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX11]], i64 0, i64 0 // CHECK5-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX11]], i64 0, i64 0
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX12]], align 4 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX12]], align 4
Show All 22 Lines
// CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK7-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 // CHECK7-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK7-NEXT: store i32 56087, ptr [[DOTOMP_UB]], align 4 // CHECK7-NEXT: store i32 56087, ptr [[DOTOMP_UB]], align 4
// CHECK7-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK7-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK7-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 // CHECK7-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK7: omp.inner.for.cond: // CHECK7: omp.inner.for.cond:
// CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3:![0-9]+]]
// CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] // CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK7: omp.inner.for.body: // CHECK7: omp.inner.for.body:
// CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 456 // CHECK7-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 456
// CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !3 // CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP5]], 456 // CHECK7-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP5]], 456
// CHECK7-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456 // CHECK7-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456
// CHECK7-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL4]] // CHECK7-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL4]]
// CHECK7-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK7-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK7-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]] // CHECK7-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK7-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group !3 // CHECK7-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 // CHECK7-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i32 0, i32 [[TMP6]] // CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i32 0, i32 [[TMP6]]
// CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP7]] // CHECK7-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP7]]
// CHECK7-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group !3 // CHECK7-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK7: omp.body.continue: // CHECK7: omp.body.continue:
// CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK7: omp.inner.for.inc: // CHECK7: omp.inner.for.inc:
// CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP8]], 1 // CHECK7-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK7-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK7-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
// CHECK7: omp.inner.for.end: // CHECK7: omp.inner.for.end:
// CHECK7-NEXT: store i32 123, ptr [[I]], align 4 // CHECK7-NEXT: store i32 123, ptr [[I]], align 4
// CHECK7-NEXT: store i32 456, ptr [[J]], align 4 // CHECK7-NEXT: store i32 456, ptr [[J]], align 4
// CHECK7-NEXT: [[A9:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 // CHECK7-NEXT: [[A9:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK7-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A9]], i32 0, i32 0 // CHECK7-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A9]], i32 0, i32 0
// CHECK7-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX10]], i32 0, i32 0 // CHECK7-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX10]], i32 0, i32 0
// CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX11]], align 4 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX11]], align 4
Show All 13 Lines
// 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 ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], 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: [[_TMP1:%.*]] = 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_2:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK9-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8
// CHECK9-NEXT: store i32 100, ptr [[N]], align 4 // CHECK9-NEXT: store i32 100, ptr [[N]], align 4
// CHECK9-NEXT: store i32 2, ptr [[M]], align 4 // CHECK9-NEXT: store i32 2, ptr [[M]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[M]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64 // CHECK9-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
// CHECK9-NEXT: [[TMP4:%.*]] = call ptr @llvm.stacksave() // CHECK9-NEXT: [[TMP4:%.*]] = call ptr @llvm.stacksave()
// CHECK9-NEXT: store ptr [[TMP4]], ptr [[SAVED_STACK]], align 8 // CHECK9-NEXT: store ptr [[TMP4]], ptr [[SAVED_STACK]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP5]], align 4 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP5]], align 4
// CHECK9-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8
// CHECK9-NEXT: store i64 [[TMP3]], ptr [[__VLA_EXPR1]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], ptr [[__VLA_EXPR1]], align 8
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[N]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP6]], ptr [[N_CASTED]], align 4 // CHECK9-NEXT: store i32 [[TMP6]], ptr [[N_CASTED]], align 4
// CHECK9-NEXT: [[TMP7:%.*]] = load i64, ptr [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP7:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[M]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP8]], ptr [[M_CASTED]], align 4 // CHECK9-NEXT: store i32 [[TMP8]], ptr [[M_CASTED]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[M_CASTED]], align 8 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[M_CASTED]], align 8
// 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: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 40, i1 false) // CHECK9-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 40, i1 false)
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP13]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP13]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP15]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP14]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP16]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store ptr null, ptr [[TMP17]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP17]], align 8
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP18]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP18]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP20]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP19]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: store ptr null, ptr [[TMP22]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP20]], align 8
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP23]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP21]], align 8
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP25]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP22]], align 8
// CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK9-NEXT: store ptr null, ptr [[TMP23]], align 8
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP24]], align 8
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP25]], align 8
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK9-NEXT: store i64 [[TMP11]], ptr [[TMP26]], align 8
// CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK9-NEXT: store ptr null, ptr [[TMP27]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP27]], align 8
// CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP28]], align 8 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP30]], align 8 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK9-NEXT: store i32 [[TMP31]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: store ptr null, ptr [[TMP32]], align 8 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK9-NEXT: store i32 [[TMP32]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP33]], align 8 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP33]], 0
// CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP35]], align 8
// CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK9-NEXT: store i64 [[TMP11]], ptr [[TMP37]], align 8
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK9-NEXT: store ptr null, ptr [[TMP38]], align 8
// CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP42:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP42]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP43:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP43]], ptr [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP44:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP44]], 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: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP45:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4 // CHECK9-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP45]], 0 // CHECK9-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP34]], 0
// CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK9-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK9-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB9]], ptr [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: store i64 [[SUB7]], ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK9-NEXT: [[TMP46:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: [[TMP35:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP46]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP35]], 1
// CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK9-NEXT: store i32 1, ptr [[TMP47]], align 4 // CHECK9-NEXT: store i32 1, ptr [[TMP36]], align 4
// CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK9-NEXT: store i32 5, ptr [[TMP48]], align 4 // CHECK9-NEXT: store i32 5, ptr [[TMP37]], align 4
// CHECK9-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK9-NEXT: store ptr [[TMP39]], ptr [[TMP49]], align 8 // CHECK9-NEXT: store ptr [[TMP28]], ptr [[TMP38]], align 8
// CHECK9-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK9-NEXT: store ptr [[TMP40]], ptr [[TMP50]], align 8 // CHECK9-NEXT: store ptr [[TMP29]], ptr [[TMP39]], align 8
// CHECK9-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK9-NEXT: store ptr [[TMP41]], ptr [[TMP51]], align 8 // CHECK9-NEXT: store ptr [[TMP30]], ptr [[TMP40]], align 8
// CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK9-NEXT: store ptr @.offload_maptypes, ptr [[TMP52]], align 8 // CHECK9-NEXT: store ptr @.offload_maptypes, ptr [[TMP41]], align 8
// CHECK9-NEXT: [[TMP53:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK9-NEXT: store ptr null, ptr [[TMP53]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP42]], align 8
// CHECK9-NEXT: [[TMP54:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK9-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK9-NEXT: store ptr null, ptr [[TMP54]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP43]], align 8
// CHECK9-NEXT: [[TMP55:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK9-NEXT: store i64 [[ADD]], ptr [[TMP55]], align 8 // CHECK9-NEXT: store i64 [[ADD]], ptr [[TMP44]], align 8
// CHECK9-NEXT: [[TMP56:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, ptr [[KERNEL_ARGS]]) // CHECK9-NEXT: [[TMP45:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, ptr [[KERNEL_ARGS]])
// CHECK9-NEXT: [[TMP57:%.*]] = icmp ne i32 [[TMP56]], 0 // CHECK9-NEXT: [[TMP46:%.*]] = icmp ne i32 [[TMP45]], 0
// CHECK9-NEXT: br i1 [[TMP57]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP46]], 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]], ptr [[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]], ptr [[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: [[TMP58:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK9-NEXT: [[TMP47:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef signext [[TMP58]]) // CHECK9-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef signext [[TMP47]])
// CHECK9-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP59:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP48:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(ptr [[TMP59]]) // CHECK9-NEXT: call void @llvm.stackrestore(ptr [[TMP48]])
// CHECK9-NEXT: [[TMP60:%.*]] = load i32, ptr [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP49:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP60]] // CHECK9-NEXT: ret i32 [[TMP49]]
// //
// //
// 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 noundef [[N:%.*]], i64 noundef [[M:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[M:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef 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 All 26 Lines
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// 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
// CHECK9-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP3:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_5:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_6:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_5:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[J:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[J:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I13:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I11:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[J14:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[J12:%.*]] = alloca i32, align 4
// 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: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8 // CHECK9-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8
// CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT: store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8 // CHECK9-NEXT: store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[CONV7:%.*]] = sext i32 [[DIV]] to i64 // CHECK9-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP6]], 0 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP6]], 0
// CHECK9-NEXT: [[DIV9:%.*]] = sdiv i32 [[SUB8]], 1 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK9-NEXT: [[CONV10:%.*]] = sext i32 [[DIV9]] to i64 // CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV7]], [[CONV10]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV8]]
// CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB11]], ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: store i64 [[SUB9]], ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: store i32 0, ptr [[I]], align 4 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4
// CHECK9-NEXT: store i32 0, ptr [[J]], align 4 // CHECK9-NEXT: store i32 0, ptr [[J]], align 4
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]] // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]] // CHECK9-NEXT: br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK9: land.lhs.true: // CHECK9: land.lhs.true:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[CMP12:%.*]] = icmp slt i32 0, [[TMP8]] // CHECK9-NEXT: [[CMP10:%.*]] = icmp slt i32 0, [[TMP8]]
// CHECK9-NEXT: br i1 [[CMP12]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]] // CHECK9-NEXT: br i1 [[CMP10]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]]
// CHECK9: omp.precond.then: // CHECK9: omp.precond.then:
// CHECK9-NEXT: store i64 0, ptr [[DOTOMP_COMB_LB]], align 8 // CHECK9-NEXT: store i64 0, ptr [[DOTOMP_COMB_LB]], align 8
// CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[DOTOMP_COMB_UB]], align 8 // CHECK9-NEXT: store i64 [[TMP9]], ptr [[DOTOMP_COMB_UB]], align 8
// CHECK9-NEXT: store i64 1, ptr [[DOTOMP_STRIDE]], align 8 // CHECK9-NEXT: store i64 1, ptr [[DOTOMP_STRIDE]], align 8
// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB1:[0-9]+]], i32 [[TMP11]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1) // CHECK9-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB1:[0-9]+]], i32 [[TMP11]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK9-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8 // CHECK9-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: [[CMP15:%.*]] = icmp sgt i64 [[TMP12]], [[TMP13]] // CHECK9-NEXT: [[CMP13:%.*]] = icmp sgt i64 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: br i1 [[CMP15]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9-NEXT: br i1 [[CMP13]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true: // CHECK9: cond.true:
// CHECK9-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false: // CHECK9: cond.false:
// CHECK9-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8 // CHECK9-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8
// CHECK9-NEXT: br label [[COND_END]] // CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end: // CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i64 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] // CHECK9-NEXT: [[COND:%.*]] = phi i64 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i64 [[COND]], ptr [[DOTOMP_COMB_UB]], align 8 // CHECK9-NEXT: store i64 [[COND]], ptr [[DOTOMP_COMB_UB]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8 // CHECK9-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
// CHECK9-NEXT: store i64 [[TMP16]], ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: store i64 [[TMP16]], ptr [[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: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !5 // CHECK9-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP5:![0-9]+]]
// CHECK9-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8, !llvm.access.group !5 // CHECK9-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[CMP16:%.*]] = icmp sle i64 [[TMP17]], [[TMP18]] // CHECK9-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP17]], [[TMP18]]
// CHECK9-NEXT: br i1 [[CMP16]], 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: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8, !llvm.access.group !5 // CHECK9-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8, !llvm.access.group !5 // CHECK9-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[N_ADDR]], align 4, !llvm.access.group !5 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[N_ADDR]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: store i32 [[TMP21]], ptr [[N_CASTED]], align 4, !llvm.access.group !5 // CHECK9-NEXT: store i32 [[TMP21]], ptr [[N_CASTED]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[TMP22:%.*]] = load i64, ptr [[N_CASTED]], align 8, !llvm.access.group !5 // CHECK9-NEXT: [[TMP22:%.*]] = load i64, ptr [[N_CASTED]], align 8, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[TMP23:%.*]] = load i32, ptr [[M_ADDR]], align 4, !llvm.access.group !5 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, ptr [[M_ADDR]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: store i32 [[TMP23]], ptr [[M_CASTED]], align 4, !llvm.access.group !5 // CHECK9-NEXT: store i32 [[TMP23]], ptr [[M_CASTED]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[TMP24:%.*]] = load i64, ptr [[M_CASTED]], align 8, !llvm.access.group !5 // CHECK9-NEXT: [[TMP24:%.*]] = load i64, ptr [[M_CASTED]], align 8, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 7, ptr @.omp_outlined..1, i64 [[TMP19]], i64 [[TMP20]], i64 [[TMP22]], i64 [[TMP24]], i64 [[TMP0]], i64 [[TMP1]], ptr [[TMP2]]), !llvm.access.group !5 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 7, ptr @.omp_outlined..1, i64 [[TMP19]], i64 [[TMP20]], i64 [[TMP22]], i64 [[TMP24]], i64 [[TMP0]], i64 [[TMP1]], ptr [[TMP2]]), !llvm.access.group [[ACC_GRP5]]
// 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: [[TMP25:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !5 // CHECK9-NEXT: [[TMP25:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[TMP26:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8, !llvm.access.group !5 // CHECK9-NEXT: [[TMP26:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP25]], [[TMP26]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP25]], [[TMP26]]
// CHECK9-NEXT: store i64 [[ADD]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group !5 // CHECK9-NEXT: store i64 [[ADD]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![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: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP28]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP28]])
// CHECK9-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0 // CHECK9-NEXT: [[TMP30:%.*]] = icmp ne i32 [[TMP29]], 0
// CHECK9-NEXT: br i1 [[TMP30]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK9-NEXT: br i1 [[TMP30]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then: // CHECK9: .omp.final.then:
// CHECK9-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB19:%.*]] = sub nsw i32 [[TMP31]], 0 // CHECK9-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP31]], 0
// CHECK9-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1
// CHECK9-NEXT: [[MUL17:%.*]] = mul nsw i32 [[DIV16]], 1
// CHECK9-NEXT: [[ADD18:%.*]] = add nsw i32 0, [[MUL17]]
// CHECK9-NEXT: store i32 [[ADD18]], ptr [[I11]], align 4
// CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[SUB19:%.*]] = sub nsw i32 [[TMP32]], 0
// CHECK9-NEXT: [[DIV20:%.*]] = sdiv i32 [[SUB19]], 1 // CHECK9-NEXT: [[DIV20:%.*]] = sdiv i32 [[SUB19]], 1
// CHECK9-NEXT: [[MUL21:%.*]] = mul nsw i32 [[DIV20]], 1 // CHECK9-NEXT: [[MUL21:%.*]] = mul nsw i32 [[DIV20]], 1
// CHECK9-NEXT: [[ADD22:%.*]] = add nsw i32 0, [[MUL21]] // CHECK9-NEXT: [[ADD22:%.*]] = add nsw i32 0, [[MUL21]]
// CHECK9-NEXT: store i32 [[ADD22]], ptr [[I13]], align 4 // CHECK9-NEXT: store i32 [[ADD22]], ptr [[J12]], align 4
// CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4
// CHECK9-NEXT: [[SUB23:%.*]] = sub nsw i32 [[TMP32]], 0
// CHECK9-NEXT: [[DIV24:%.*]] = sdiv i32 [[SUB23]], 1
// CHECK9-NEXT: [[MUL25:%.*]] = mul nsw i32 [[DIV24]], 1
// CHECK9-NEXT: [[ADD26:%.*]] = add nsw i32 0, [[MUL25]]
// CHECK9-NEXT: store i32 [[ADD26]], ptr [[J14]], 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..1 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i64 noundef [[N:%.*]], i64 noundef [[M:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i64 noundef [[N:%.*]], i64 noundef [[M:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, 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: [[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
// CHECK9-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP3:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_5:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_6:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_5:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[J:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[J:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I13:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I11:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[J14:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[J12:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8 // CHECK9-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8
// CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT: store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8 // CHECK9-NEXT: store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[CONV7:%.*]] = sext i32 [[DIV]] to i64 // CHECK9-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP6]], 0 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP6]], 0
// CHECK9-NEXT: [[DIV9:%.*]] = sdiv i32 [[SUB8]], 1 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK9-NEXT: [[CONV10:%.*]] = sext i32 [[DIV9]] to i64 // CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV7]], [[CONV10]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV8]]
// CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB11]], ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: store i64 [[SUB9]], ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: store i32 0, ptr [[I]], align 4 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4
// CHECK9-NEXT: store i32 0, ptr [[J]], align 4 // CHECK9-NEXT: store i32 0, ptr [[J]], align 4
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]] // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]] // CHECK9-NEXT: br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK9: land.lhs.true: // CHECK9: land.lhs.true:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[CMP12:%.*]] = icmp slt i32 0, [[TMP8]] // CHECK9-NEXT: [[CMP10:%.*]] = icmp slt i32 0, [[TMP8]]
// CHECK9-NEXT: br i1 [[CMP12]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]] // CHECK9-NEXT: br i1 [[CMP10]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]]
// CHECK9: omp.precond.then: // CHECK9: omp.precond.then:
// CHECK9-NEXT: store i64 0, ptr [[DOTOMP_LB]], align 8 // CHECK9-NEXT: store i64 0, ptr [[DOTOMP_LB]], align 8
// CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: store i64 [[TMP9]], ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8 // CHECK9-NEXT: [[TMP10:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 // CHECK9-NEXT: [[TMP11:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8
// CHECK9-NEXT: store i64 [[TMP10]], ptr [[DOTOMP_LB]], align 8 // CHECK9-NEXT: store i64 [[TMP10]], ptr [[DOTOMP_LB]], align 8
// CHECK9-NEXT: store i64 [[TMP11]], ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: store i64 [[TMP11]], ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: store i64 1, ptr [[DOTOMP_STRIDE]], align 8 // CHECK9-NEXT: store i64 1, ptr [[DOTOMP_STRIDE]], align 8
// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB2:[0-9]+]], i32 [[TMP13]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1) // CHECK9-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB2:[0-9]+]], i32 [[TMP13]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK9-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: [[CMP15:%.*]] = icmp sgt i64 [[TMP14]], [[TMP15]] // CHECK9-NEXT: [[CMP13:%.*]] = icmp sgt i64 [[TMP14]], [[TMP15]]
// CHECK9-NEXT: br i1 [[CMP15]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9-NEXT: br i1 [[CMP13]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true: // CHECK9: cond.true:
// CHECK9-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false: // CHECK9: cond.false:
// CHECK9-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// 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]], ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: store i64 [[COND]], ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8 // CHECK9-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK9-NEXT: store i64 [[TMP18]], ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: store i64 [[TMP18]], ptr [[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, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !9 // CHECK9-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP9:![0-9]+]]
// CHECK9-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group !9 // CHECK9-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group [[ACC_GRP9]]
// CHECK9-NEXT: [[CMP16:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]] // CHECK9-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]]
// CHECK9-NEXT: br i1 [[CMP16]], 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, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !9 // CHECK9-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP9]]
// CHECK9-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4, !llvm.access.group !9 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK9-NEXT: [[SUB17:%.*]] = sub nsw i32 [[TMP22]], 0 // CHECK9-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP22]], 0
// CHECK9-NEXT: [[DIV18:%.*]] = sdiv i32 [[SUB17]], 1 // CHECK9-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1
// CHECK9-NEXT: [[MUL19:%.*]] = mul nsw i32 1, [[DIV18]] // CHECK9-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]]
// CHECK9-NEXT: [[CONV20:%.*]] = sext i32 [[MUL19]] to i64 // CHECK9-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64
// CHECK9-NEXT: [[DIV21:%.*]] = sdiv i64 [[TMP21]], [[CONV20]] // CHECK9-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP21]], [[CONV18]]
// CHECK9-NEXT: [[MUL22:%.*]] = mul nsw i64 [[DIV21]], 1 // CHECK9-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL22]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]]
// CHECK9-NEXT: [[CONV23:%.*]] = trunc i64 [[ADD]] to i32 // CHECK9-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32
// CHECK9-NEXT: store i32 [[CONV23]], ptr [[I13]], align 4, !llvm.access.group !9 // CHECK9-NEXT: store i32 [[CONV21]], ptr [[I11]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK9-NEXT: [[TMP23:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !9 // CHECK9-NEXT: [[TMP23:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP9]]
// CHECK9-NEXT: [[TMP24:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !9 // CHECK9-NEXT: [[TMP24:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP9]]
// CHECK9-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4, !llvm.access.group !9 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK9-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP25]], 0 // CHECK9-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP25]], 0
// CHECK9-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1 // CHECK9-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1
// CHECK9-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]] // CHECK9-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]]
// CHECK9-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64 // CHECK9-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64
// CHECK9-NEXT: [[DIV28:%.*]] = sdiv i64 [[TMP24]], [[CONV27]] // CHECK9-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP24]], [[CONV25]]
// CHECK9-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4, !llvm.access.group !9 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK9-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP26]], 0 // CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK9-NEXT: [[DIV30:%.*]] = sdiv i32 [[SUB29]], 1 // CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK9-NEXT: [[MUL31:%.*]] = mul nsw i32 1, [[DIV30]] // CHECK9-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]]
// CHECK9-NEXT: [[CONV32:%.*]] = sext i32 [[MUL31]] to i64 // CHECK9-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64
// CHECK9-NEXT: [[MUL33:%.*]] = mul nsw i64 [[DIV28]], [[CONV32]] // CHECK9-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]]
// CHECK9-NEXT: [[SUB34:%.*]] = sub nsw i64 [[TMP23]], [[MUL33]] // CHECK9-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP23]], [[MUL31]]
// CHECK9-NEXT: [[MUL35:%.*]] = mul nsw i64 [[SUB34]], 1 // CHECK9-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1
// CHECK9-NEXT: [[ADD36:%.*]] = add nsw i64 0, [[MUL35]] // CHECK9-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]]
// CHECK9-NEXT: [[CONV37:%.*]] = trunc i64 [[ADD36]] to i32 // CHECK9-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32
// CHECK9-NEXT: store i32 [[CONV37]], ptr [[J14]], align 4, !llvm.access.group !9 // CHECK9-NEXT: store i32 [[CONV35]], ptr [[J12]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, ptr [[I13]], align 4, !llvm.access.group !9 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, ptr [[I11]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP27]] to i64 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP27]] to i64
// CHECK9-NEXT: [[TMP28:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP1]] // CHECK9-NEXT: [[TMP28:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP1]]
// CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[TMP28]] // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[TMP28]]
// CHECK9-NEXT: [[TMP29:%.*]] = load i32, ptr [[J14]], align 4, !llvm.access.group !9 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, ptr [[J12]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK9-NEXT: [[IDXPROM38:%.*]] = sext i32 [[TMP29]] to i64 // CHECK9-NEXT: [[IDXPROM36:%.*]] = sext i32 [[TMP29]] to i64
// CHECK9-NEXT: [[ARRAYIDX39:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i64 [[IDXPROM38]] // CHECK9-NEXT: [[ARRAYIDX37:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i64 [[IDXPROM36]]
// CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX39]], align 4, !llvm.access.group !9 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX37]], align 4, !llvm.access.group [[ACC_GRP9]]
// 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, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !9 // CHECK9-NEXT: [[TMP30:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP9]]
// CHECK9-NEXT: [[ADD40:%.*]] = add nsw i64 [[TMP30]], 1 // CHECK9-NEXT: [[ADD38:%.*]] = add nsw i64 [[TMP30]], 1
// CHECK9-NEXT: store i64 [[ADD40]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group !9 // CHECK9-NEXT: store i64 [[ADD38]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP9]]
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![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: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP32]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP32]])
// CHECK9-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0 // CHECK9-NEXT: [[TMP34:%.*]] = icmp ne i32 [[TMP33]], 0
// CHECK9-NEXT: br i1 [[TMP34]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK9-NEXT: br i1 [[TMP34]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then: // CHECK9: .omp.final.then:
// CHECK9-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP35:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB41:%.*]] = sub nsw i32 [[TMP35]], 0 // CHECK9-NEXT: [[SUB39:%.*]] = sub nsw i32 [[TMP35]], 0
// CHECK9-NEXT: [[DIV42:%.*]] = sdiv i32 [[SUB41]], 1 // CHECK9-NEXT: [[DIV40:%.*]] = sdiv i32 [[SUB39]], 1
// CHECK9-NEXT: [[MUL43:%.*]] = mul nsw i32 [[DIV42]], 1 // CHECK9-NEXT: [[MUL41:%.*]] = mul nsw i32 [[DIV40]], 1
// CHECK9-NEXT: [[ADD44:%.*]] = add nsw i32 0, [[MUL43]] // CHECK9-NEXT: [[ADD42:%.*]] = add nsw i32 0, [[MUL41]]
// CHECK9-NEXT: store i32 [[ADD44]], ptr [[I13]], align 4 // CHECK9-NEXT: store i32 [[ADD42]], ptr [[I11]], align 4
// CHECK9-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP36:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[SUB45:%.*]] = sub nsw i32 [[TMP36]], 0 // CHECK9-NEXT: [[SUB43:%.*]] = sub nsw i32 [[TMP36]], 0
// CHECK9-NEXT: [[DIV46:%.*]] = sdiv i32 [[SUB45]], 1 // CHECK9-NEXT: [[DIV44:%.*]] = sdiv i32 [[SUB43]], 1
// CHECK9-NEXT: [[MUL47:%.*]] = mul nsw i32 [[DIV46]], 1 // CHECK9-NEXT: [[MUL45:%.*]] = mul nsw i32 [[DIV44]], 1
// CHECK9-NEXT: [[ADD48:%.*]] = add nsw i32 0, [[MUL47]] // CHECK9-NEXT: [[ADD46:%.*]] = add nsw i32 0, [[MUL45]]
// CHECK9-NEXT: store i32 [[ADD48]], ptr [[J14]], align 4 // CHECK9-NEXT: store i32 [[ADD46]], ptr [[J12]], 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 {{[^@]+}}@_Z5tmainIiLi10ELi2EEiT_ // CHECK9-LABEL: define {{[^@]+}}@_Z5tmainIiLi10ELi2EEiT_
// CHECK9-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { // CHECK9-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[A:%.*]] = alloca [10 x [2 x i32]], align 4 // CHECK9-NEXT: [[A:%.*]] = alloca [10 x [2 x i32]], align 4
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store ptr [[A]], ptr [[TMP0]], align 8 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP0]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: store ptr [[A]], ptr [[TMP2]], align 8 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP1]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK9-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK9-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK9-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK9-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK9-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8 // CHECK9-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK9-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK9-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK9-NEXT: store ptr @.offload_sizes.4, ptr [[TMP11]], align 8 // CHECK9-NEXT: store ptr @.offload_sizes.4, ptr [[TMP9]], align 8
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK9-NEXT: store ptr @.offload_maptypes.5, ptr [[TMP12]], align 8 // CHECK9-NEXT: store ptr @.offload_maptypes.5, ptr [[TMP10]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK9-NEXT: store ptr null, ptr [[TMP13]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK9-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK9-NEXT: store i64 20, ptr [[TMP15]], align 8 // CHECK9-NEXT: store i64 20, ptr [[TMP13]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, ptr [[KERNEL_ARGS]]) // CHECK9-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, ptr [[KERNEL_ARGS]])
// CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP15]], 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(ptr [[A]]) #[[ATTR3]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68(ptr [[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
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68
▲ Show 20 Lines • Show All 42 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]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[TMP5]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !14 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14:![0-9]+]]
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !14 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP14]]
// 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, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !14 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP14]]
// CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !14 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP14]]
// CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @.omp_outlined..3, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]]), !llvm.access.group !14 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @.omp_outlined..3, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]]), !llvm.access.group [[ACC_GRP14]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !14 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14]]
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group !14 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP14]]
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !14 // CHECK9-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14]]
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[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:%.*]]
▲ Show 20 Lines • Show All 51 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]], ptr [[DOTOMP_UB]], align 4 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[TMP7]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !17 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17:![0-9]+]]
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !17 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP17]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !17 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
// 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]], ptr [[I]], align 4, !llvm.access.group !17 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !17 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !17 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
// 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]], ptr [[J]], align 4, !llvm.access.group !17 // CHECK9-NEXT: store i32 [[ADD7]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !17 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP17]]
// 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]], ptr [[TMP0]], i64 0, i64 [[IDXPROM]] // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
// CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group !17 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK9-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP14]] to i64 // CHECK9-NEXT: [[IDXPROM8:%.*]] = sext i32 [[TMP14]] to i64
// CHECK9-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM8]] // CHECK9-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM8]]
// CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX9]], align 4, !llvm.access.group !17 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX9]], align 4, !llvm.access.group [[ACC_GRP17]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !17 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK9-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP15]], 1 // CHECK9-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK9-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !17 // CHECK9-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP18:![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(ptr @[[GLOB1]], i32 [[TMP4]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP4]])
// CHECK9-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, ptr [[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:%.*]]
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[M]], align 4 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP6]], ptr [[M_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP6]], ptr [[M_CASTED]], align 4
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[M_CASTED]], align 4 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[M_CASTED]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP0]], [[TMP1]] // CHECK11-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP0]], [[TMP1]]
// 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: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 40, i1 false) // CHECK11-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 40, i1 false)
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP11]], align 4
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP12]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP12]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP14]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP13]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store ptr null, ptr [[TMP16]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP16]], align 4
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP17]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP17]], align 4
// CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP19]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP18]], align 4
// CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: store ptr null, ptr [[TMP21]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP19]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP22]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP20]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP24]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP21]], align 4
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK11-NEXT: store ptr null, ptr [[TMP22]], align 4
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP23]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP24]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK11-NEXT: store i64 [[TMP10]], ptr [[TMP25]], align 4
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK11-NEXT: store ptr null, ptr [[TMP26]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP26]], align 4
// CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP27]], align 4 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP29]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK11-NEXT: store i32 [[TMP30]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store ptr null, ptr [[TMP31]], align 4 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, ptr [[M]], align 4
// CHECK11-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK11-NEXT: store i32 [[TMP31]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP32]], align 4 // CHECK11-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP32]], 0
// CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP34]], align 4
// CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK11-NEXT: store i64 [[TMP10]], ptr [[TMP36]], align 4
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK11-NEXT: store ptr null, ptr [[TMP37]], align 4
// CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP41:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP41]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = load i32, ptr [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP42]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP43:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP43]], 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: [[TMP44:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP44]], 0 // CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP33]], 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]], ptr [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: store i64 [[SUB7]], ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[TMP45:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: [[TMP34:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP45]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP34]], 1
// CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK11-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK11-NEXT: store i32 1, ptr [[TMP46]], align 4 // CHECK11-NEXT: store i32 1, ptr [[TMP35]], align 4
// CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK11-NEXT: store i32 5, ptr [[TMP47]], align 4 // CHECK11-NEXT: store i32 5, ptr [[TMP36]], align 4
// CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK11-NEXT: store ptr [[TMP38]], ptr [[TMP48]], align 4 // CHECK11-NEXT: store ptr [[TMP27]], ptr [[TMP37]], align 4
// CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK11-NEXT: store ptr [[TMP39]], ptr [[TMP49]], align 4 // CHECK11-NEXT: store ptr [[TMP28]], ptr [[TMP38]], align 4
// CHECK11-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK11-NEXT: store ptr [[TMP40]], ptr [[TMP50]], align 4 // CHECK11-NEXT: store ptr [[TMP29]], ptr [[TMP39]], align 4
// CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK11-NEXT: store ptr @.offload_maptypes, ptr [[TMP51]], align 4 // CHECK11-NEXT: store ptr @.offload_maptypes, ptr [[TMP40]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK11-NEXT: store ptr null, ptr [[TMP52]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP41]], align 4
// CHECK11-NEXT: [[TMP53:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK11-NEXT: store ptr null, ptr [[TMP53]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP42]], align 4
// CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK11-NEXT: store i64 [[ADD]], ptr [[TMP54]], align 8 // CHECK11-NEXT: store i64 [[ADD]], ptr [[TMP43]], align 8
// CHECK11-NEXT: [[TMP55:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, ptr [[KERNEL_ARGS]]) // CHECK11-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l81.region_id, ptr [[KERNEL_ARGS]])
// CHECK11-NEXT: [[TMP56:%.*]] = icmp ne i32 [[TMP55]], 0 // CHECK11-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK11-NEXT: br i1 [[TMP56]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP45]], 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]], ptr [[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]], ptr [[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: [[TMP57:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK11-NEXT: [[TMP46:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef [[TMP57]]) // CHECK11-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef [[TMP46]])
// CHECK11-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP58:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(ptr [[TMP58]]) // CHECK11-NEXT: call void @llvm.stackrestore(ptr [[TMP47]])
// CHECK11-NEXT: [[TMP59:%.*]] = load i32, ptr [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP48:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP59]] // CHECK11-NEXT: ret i32 [[TMP48]]
// //
// //
// 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 noundef [[N:%.*]], i32 noundef [[M:%.*]], i32 noundef [[VLA:%.*]], i32 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[M:%.*]], i32 noundef [[VLA:%.*]], i32 noundef [[VLA1:%.*]], ptr noundef 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 100 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 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] // CHECK11-NEXT: [[COND:%.*]] = phi i64 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ]
// CHECK11-NEXT: store i64 [[COND]], ptr [[DOTOMP_COMB_UB]], align 8 // CHECK11-NEXT: store i64 [[COND]], ptr [[DOTOMP_COMB_UB]], align 8
// CHECK11-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8 // CHECK11-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8
// CHECK11-NEXT: store i64 [[TMP16]], ptr [[DOTOMP_IV]], align 8 // CHECK11-NEXT: store i64 [[TMP16]], ptr [[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: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !6 // CHECK11-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP6:![0-9]+]]
// CHECK11-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8, !llvm.access.group !6 // CHECK11-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP17]], [[TMP18]] // CHECK11-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP17]], [[TMP18]]
// 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: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8, !llvm.access.group !6 // CHECK11-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_COMB_LB]], align 8, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[TMP20:%.*]] = trunc i64 [[TMP19]] to i32 // CHECK11-NEXT: [[TMP20:%.*]] = trunc i64 [[TMP19]] to i32
// CHECK11-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8, !llvm.access.group !6 // CHECK11-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTOMP_COMB_UB]], align 8, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[TMP22:%.*]] = trunc i64 [[TMP21]] to i32 // CHECK11-NEXT: [[TMP22:%.*]] = trunc i64 [[TMP21]] to i32
// CHECK11-NEXT: [[TMP23:%.*]] = load i32, ptr [[N_ADDR]], align 4, !llvm.access.group !6 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, ptr [[N_ADDR]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: store i32 [[TMP23]], ptr [[N_CASTED]], align 4, !llvm.access.group !6 // CHECK11-NEXT: store i32 [[TMP23]], ptr [[N_CASTED]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[TMP24:%.*]] = load i32, ptr [[N_CASTED]], align 4, !llvm.access.group !6 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, ptr [[N_CASTED]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[TMP25:%.*]] = load i32, ptr [[M_ADDR]], align 4, !llvm.access.group !6 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, ptr [[M_ADDR]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: store i32 [[TMP25]], ptr [[M_CASTED]], align 4, !llvm.access.group !6 // CHECK11-NEXT: store i32 [[TMP25]], ptr [[M_CASTED]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[TMP26:%.*]] = load i32, ptr [[M_CASTED]], align 4, !llvm.access.group !6 // CHECK11-NEXT: [[TMP26:%.*]] = load i32, ptr [[M_CASTED]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 7, ptr @.omp_outlined..1, i32 [[TMP20]], i32 [[TMP22]], i32 [[TMP24]], i32 [[TMP26]], i32 [[TMP0]], i32 [[TMP1]], ptr [[TMP2]]), !llvm.access.group !6 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 7, ptr @.omp_outlined..1, i32 [[TMP20]], i32 [[TMP22]], i32 [[TMP24]], i32 [[TMP26]], i32 [[TMP0]], i32 [[TMP1]], ptr [[TMP2]]), !llvm.access.group [[ACC_GRP6]]
// 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: [[TMP27:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !6 // CHECK11-NEXT: [[TMP27:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[TMP28:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8, !llvm.access.group !6 // CHECK11-NEXT: [[TMP28:%.*]] = load i64, ptr [[DOTOMP_STRIDE]], align 8, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP27]], [[TMP28]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP27]], [[TMP28]]
// CHECK11-NEXT: store i64 [[ADD]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group !6 // CHECK11-NEXT: store i64 [[ADD]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![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: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4
// CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP30]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP30]])
// CHECK11-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
▲ Show 20 Lines • Show All 108 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]], ptr [[DOTOMP_UB]], align 8 // CHECK11-NEXT: store i64 [[COND]], ptr [[DOTOMP_UB]], align 8
// CHECK11-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8 // CHECK11-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK11-NEXT: store i64 [[TMP18]], ptr [[DOTOMP_IV]], align 8 // CHECK11-NEXT: store i64 [[TMP18]], ptr [[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, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !10 // CHECK11-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP10:![0-9]+]]
// CHECK11-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group !10 // CHECK11-NEXT: [[TMP20:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group [[ACC_GRP10]]
// CHECK11-NEXT: [[CMP16:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]] // CHECK11-NEXT: [[CMP16:%.*]] = icmp sle i64 [[TMP19]], [[TMP20]]
// 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: [[TMP21:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !10 // CHECK11-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP10]]
// CHECK11-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !10 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP10]]
// CHECK11-NEXT: [[SUB17:%.*]] = sub nsw i32 [[TMP22]], 0 // CHECK11-NEXT: [[SUB17:%.*]] = sub nsw i32 [[TMP22]], 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 [[TMP21]], [[CONV20]] // CHECK11-NEXT: [[DIV21:%.*]] = sdiv i64 [[TMP21]], [[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]], ptr [[I13]], align 4, !llvm.access.group !10 // CHECK11-NEXT: store i32 [[CONV23]], ptr [[I13]], align 4, !llvm.access.group [[ACC_GRP10]]
// CHECK11-NEXT: [[TMP23:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !10 // CHECK11-NEXT: [[TMP23:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP10]]
// CHECK11-NEXT: [[TMP24:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !10 // CHECK11-NEXT: [[TMP24:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP10]]
// CHECK11-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !10 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP10]]
// CHECK11-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP25]], 0 // CHECK11-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP25]], 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 [[TMP24]], [[CONV27]] // CHECK11-NEXT: [[DIV28:%.*]] = sdiv i64 [[TMP24]], [[CONV27]]
// CHECK11-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !10 // CHECK11-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP10]]
// CHECK11-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP26]], 0 // CHECK11-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP26]], 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 [[TMP23]], [[MUL33]] // CHECK11-NEXT: [[SUB34:%.*]] = sub nsw i64 [[TMP23]], [[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]], ptr [[J14]], align 4, !llvm.access.group !10 // CHECK11-NEXT: store i32 [[CONV37]], ptr [[J14]], align 4, !llvm.access.group [[ACC_GRP10]]
// CHECK11-NEXT: [[TMP27:%.*]] = load i32, ptr [[I13]], align 4, !llvm.access.group !10 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, ptr [[I13]], align 4, !llvm.access.group [[ACC_GRP10]]
// CHECK11-NEXT: [[TMP28:%.*]] = mul nsw i32 [[TMP27]], [[TMP1]] // CHECK11-NEXT: [[TMP28:%.*]] = mul nsw i32 [[TMP27]], [[TMP1]]
// CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP28]] // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP28]]
// CHECK11-NEXT: [[TMP29:%.*]] = load i32, ptr [[J14]], align 4, !llvm.access.group !10 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, ptr [[J14]], align 4, !llvm.access.group [[ACC_GRP10]]
// CHECK11-NEXT: [[ARRAYIDX38:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i32 [[TMP29]] // CHECK11-NEXT: [[ARRAYIDX38:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i32 [[TMP29]]
// CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX38]], align 4, !llvm.access.group !10 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX38]], align 4, !llvm.access.group [[ACC_GRP10]]
// 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, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !10 // CHECK11-NEXT: [[TMP30:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP10]]
// CHECK11-NEXT: [[ADD39:%.*]] = add nsw i64 [[TMP30]], 1 // CHECK11-NEXT: [[ADD39:%.*]] = add nsw i64 [[TMP30]], 1
// CHECK11-NEXT: store i64 [[ADD39]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group !10 // CHECK11-NEXT: store i64 [[ADD39]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP10]]
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP11:![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: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK11-NEXT: [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4 // CHECK11-NEXT: [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
// CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP32]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP32]])
// CHECK11-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
Show All 27 Lines
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], 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: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store ptr [[A]], ptr [[TMP0]], align 4 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP0]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: store ptr [[A]], ptr [[TMP2]], align 4 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP1]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK11-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK11-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK11-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK11-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK11-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 // CHECK11-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK11-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK11-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK11-NEXT: store ptr @.offload_sizes.4, ptr [[TMP11]], align 4 // CHECK11-NEXT: store ptr @.offload_sizes.4, ptr [[TMP9]], align 4
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK11-NEXT: store ptr @.offload_maptypes.5, ptr [[TMP12]], align 4 // CHECK11-NEXT: store ptr @.offload_maptypes.5, ptr [[TMP10]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK11-NEXT: store ptr null, ptr [[TMP13]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK11-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK11-NEXT: store i64 20, ptr [[TMP15]], align 8 // CHECK11-NEXT: store i64 20, ptr [[TMP13]], align 8
// CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, ptr [[KERNEL_ARGS]]) // CHECK11-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68.region_id, ptr [[KERNEL_ARGS]])
// CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK11-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP15]], 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(ptr [[A]]) #[[ATTR3]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68(ptr [[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
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l68
▲ Show 20 Lines • Show All 42 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]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK11-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !15 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15:![0-9]+]]
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !15 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP15]]
// 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, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !15 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP15]]
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !15 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP15]]
// CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @.omp_outlined..3, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]]), !llvm.access.group !15 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @.omp_outlined..3, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]]), !llvm.access.group [[ACC_GRP15]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !15 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15]]
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group !15 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP15]]
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK11-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !15 // CHECK11-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15]]
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[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:%.*]]
▲ Show 20 Lines • Show All 49 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]], ptr [[DOTOMP_UB]], align 4 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 // CHECK11-NEXT: store i32 [[TMP7]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !18 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18:![0-9]+]]
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !18 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP18]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !18 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18]]
// 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]], ptr [[I]], align 4, !llvm.access.group !18 // CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP18]]
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !18 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18]]
// CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !18 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18]]
// 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]], ptr [[J]], align 4, !llvm.access.group !18 // CHECK11-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP18]]
// CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !18 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP18]]
// CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[TMP0]], i32 0, i32 [[TMP13]] // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[TMP0]], i32 0, i32 [[TMP13]]
// CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group !18 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP18]]
// CHECK11-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP14]] // CHECK11-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP14]]
// CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group !18 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group [[ACC_GRP18]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !18 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18]]
// CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP15]], 1 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK11-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !18 // CHECK11-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18]]
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![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(ptr @[[GLOB1]], i32 [[TMP4]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP4]])
// CHECK11-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, ptr [[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:%.*]]
▲ Show 20 Lines • Show All 77 Lines • ▼ Show 20 Lines
// CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK13-NEXT: [[CMP8:%.*]] = icmp slt i32 0, [[TMP12]] // CHECK13-NEXT: [[CMP8:%.*]] = icmp slt i32 0, [[TMP12]]
// CHECK13-NEXT: br i1 [[CMP8]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END]] // CHECK13-NEXT: br i1 [[CMP8]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END]]
// CHECK13: simd.if.then: // CHECK13: simd.if.then:
// CHECK13-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8 // CHECK13-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK13-NEXT: store i64 [[TMP13]], ptr [[DOTOMP_IV]], align 8 // CHECK13-NEXT: store i64 [[TMP13]], ptr [[DOTOMP_IV]], align 8
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK13: omp.inner.for.cond: // CHECK13: omp.inner.for.cond:
// CHECK13-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !2 // CHECK13-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2:![0-9]+]]
// CHECK13-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group !2 // CHECK13-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[CMP11:%.*]] = icmp sle i64 [[TMP14]], [[TMP15]] // CHECK13-NEXT: [[CMP11:%.*]] = icmp sle i64 [[TMP14]], [[TMP15]]
// CHECK13-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK13: omp.inner.for.body: // CHECK13: omp.inner.for.body:
// CHECK13-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !2 // CHECK13-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group !2 // CHECK13-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP17]], 0 // CHECK13-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP17]], 0
// CHECK13-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK13-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1
// CHECK13-NEXT: [[MUL14:%.*]] = mul nsw i32 1, [[DIV13]] // CHECK13-NEXT: [[MUL14:%.*]] = mul nsw i32 1, [[DIV13]]
// CHECK13-NEXT: [[CONV15:%.*]] = sext i32 [[MUL14]] to i64 // CHECK13-NEXT: [[CONV15:%.*]] = sext i32 [[MUL14]] to i64
// CHECK13-NEXT: [[DIV16:%.*]] = sdiv i64 [[TMP16]], [[CONV15]] // CHECK13-NEXT: [[DIV16:%.*]] = sdiv i64 [[TMP16]], [[CONV15]]
// CHECK13-NEXT: [[MUL17:%.*]] = mul nsw i64 [[DIV16]], 1 // CHECK13-NEXT: [[MUL17:%.*]] = mul nsw i64 [[DIV16]], 1
// CHECK13-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL17]] // CHECK13-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL17]]
// CHECK13-NEXT: [[CONV18:%.*]] = trunc i64 [[ADD]] to i32 // CHECK13-NEXT: [[CONV18:%.*]] = trunc i64 [[ADD]] to i32
// CHECK13-NEXT: store i32 [[CONV18]], ptr [[I9]], align 4, !llvm.access.group !2 // CHECK13-NEXT: store i32 [[CONV18]], ptr [[I9]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !2 // CHECK13-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !2 // CHECK13-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group !2 // CHECK13-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[SUB19:%.*]] = sub nsw i32 [[TMP20]], 0 // CHECK13-NEXT: [[SUB19:%.*]] = sub nsw i32 [[TMP20]], 0
// CHECK13-NEXT: [[DIV20:%.*]] = sdiv i32 [[SUB19]], 1 // CHECK13-NEXT: [[DIV20:%.*]] = sdiv i32 [[SUB19]], 1
// CHECK13-NEXT: [[MUL21:%.*]] = mul nsw i32 1, [[DIV20]] // CHECK13-NEXT: [[MUL21:%.*]] = mul nsw i32 1, [[DIV20]]
// CHECK13-NEXT: [[CONV22:%.*]] = sext i32 [[MUL21]] to i64 // CHECK13-NEXT: [[CONV22:%.*]] = sext i32 [[MUL21]] to i64
// CHECK13-NEXT: [[DIV23:%.*]] = sdiv i64 [[TMP19]], [[CONV22]] // CHECK13-NEXT: [[DIV23:%.*]] = sdiv i64 [[TMP19]], [[CONV22]]
// CHECK13-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group !2 // CHECK13-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP21]], 0 // CHECK13-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP21]], 0
// CHECK13-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1 // CHECK13-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1
// CHECK13-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]] // CHECK13-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]]
// CHECK13-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64 // CHECK13-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64
// CHECK13-NEXT: [[MUL28:%.*]] = mul nsw i64 [[DIV23]], [[CONV27]] // CHECK13-NEXT: [[MUL28:%.*]] = mul nsw i64 [[DIV23]], [[CONV27]]
// CHECK13-NEXT: [[SUB29:%.*]] = sub nsw i64 [[TMP18]], [[MUL28]] // CHECK13-NEXT: [[SUB29:%.*]] = sub nsw i64 [[TMP18]], [[MUL28]]
// CHECK13-NEXT: [[MUL30:%.*]] = mul nsw i64 [[SUB29]], 1 // CHECK13-NEXT: [[MUL30:%.*]] = mul nsw i64 [[SUB29]], 1
// CHECK13-NEXT: [[ADD31:%.*]] = add nsw i64 0, [[MUL30]] // CHECK13-NEXT: [[ADD31:%.*]] = add nsw i64 0, [[MUL30]]
// CHECK13-NEXT: [[CONV32:%.*]] = trunc i64 [[ADD31]] to i32 // CHECK13-NEXT: [[CONV32:%.*]] = trunc i64 [[ADD31]] to i32
// CHECK13-NEXT: store i32 [[CONV32]], ptr [[J10]], align 4, !llvm.access.group !2 // CHECK13-NEXT: store i32 [[CONV32]], ptr [[J10]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[TMP22:%.*]] = load i32, ptr [[I9]], align 4, !llvm.access.group !2 // CHECK13-NEXT: [[TMP22:%.*]] = load i32, ptr [[I9]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP22]] to i64 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP22]] to i64
// CHECK13-NEXT: [[TMP23:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP3]] // CHECK13-NEXT: [[TMP23:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP3]]
// CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i64 [[TMP23]] // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i64 [[TMP23]]
// CHECK13-NEXT: [[TMP24:%.*]] = load i32, ptr [[J10]], align 4, !llvm.access.group !2 // CHECK13-NEXT: [[TMP24:%.*]] = load i32, ptr [[J10]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[IDXPROM33:%.*]] = sext i32 [[TMP24]] to i64 // CHECK13-NEXT: [[IDXPROM33:%.*]] = sext i32 [[TMP24]] to i64
// CHECK13-NEXT: [[ARRAYIDX34:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i64 [[IDXPROM33]] // CHECK13-NEXT: [[ARRAYIDX34:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i64 [[IDXPROM33]]
// CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX34]], align 4, !llvm.access.group !2 // CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX34]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK13: omp.body.continue: // CHECK13: omp.body.continue:
// CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK13: omp.inner.for.inc: // CHECK13: omp.inner.for.inc:
// CHECK13-NEXT: [[TMP25:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !2 // CHECK13-NEXT: [[TMP25:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[ADD35:%.*]] = add nsw i64 [[TMP25]], 1 // CHECK13-NEXT: [[ADD35:%.*]] = add nsw i64 [[TMP25]], 1
// CHECK13-NEXT: store i64 [[ADD35]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group !2 // CHECK13-NEXT: store i64 [[ADD35]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK13: omp.inner.for.end: // CHECK13: omp.inner.for.end:
// CHECK13-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK13-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK13-NEXT: [[SUB36:%.*]] = sub nsw i32 [[TMP26]], 0 // CHECK13-NEXT: [[SUB36:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK13-NEXT: [[DIV37:%.*]] = sdiv i32 [[SUB36]], 1 // CHECK13-NEXT: [[DIV37:%.*]] = sdiv i32 [[SUB36]], 1
// CHECK13-NEXT: [[MUL38:%.*]] = mul nsw i32 [[DIV37]], 1 // CHECK13-NEXT: [[MUL38:%.*]] = mul nsw i32 [[DIV37]], 1
// CHECK13-NEXT: [[ADD39:%.*]] = add nsw i32 0, [[MUL38]] // CHECK13-NEXT: [[ADD39:%.*]] = add nsw i32 0, [[MUL38]]
// CHECK13-NEXT: store i32 [[ADD39]], ptr [[I9]], align 4 // CHECK13-NEXT: store i32 [[ADD39]], ptr [[I9]], align 4
Show All 28 Lines
// CHECK13-NEXT: [[J:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[J:%.*]] = alloca i32, align 4
// CHECK13-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK13-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK13-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 19, ptr [[DOTOMP_UB]], align 4 // CHECK13-NEXT: store i32 19, ptr [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK13-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 // CHECK13-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK13: omp.inner.for.cond: // CHECK13: omp.inner.for.cond:
// CHECK13-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK13-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6:![0-9]+]]
// CHECK13-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !6 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] // CHECK13-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK13-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK13: omp.inner.for.body: // CHECK13: omp.inner.for.body:
// CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 2 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 2
// CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK13-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !6 // CHECK13-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[DIV2:%.*]] = sdiv i32 [[TMP5]], 2 // CHECK13-NEXT: [[DIV2:%.*]] = sdiv i32 [[TMP5]], 2
// CHECK13-NEXT: [[MUL3:%.*]] = mul nsw i32 [[DIV2]], 2 // CHECK13-NEXT: [[MUL3:%.*]] = mul nsw i32 [[DIV2]], 2
// CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL3]] // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL3]]
// CHECK13-NEXT: [[MUL4:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK13-NEXT: [[MUL4:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK13-NEXT: [[ADD5:%.*]] = add nsw i32 0, [[MUL4]] // CHECK13-NEXT: [[ADD5:%.*]] = add nsw i32 0, [[MUL4]]
// CHECK13-NEXT: store i32 [[ADD5]], ptr [[J]], align 4, !llvm.access.group !6 // CHECK13-NEXT: store i32 [[ADD5]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !6 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP6]] to i64 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP6]] to i64
// CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[A]], i64 0, i64 [[IDXPROM]] // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[A]], i64 0, i64 [[IDXPROM]]
// CHECK13-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group !6 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[IDXPROM6:%.*]] = sext i32 [[TMP7]] to i64 // CHECK13-NEXT: [[IDXPROM6:%.*]] = sext i32 [[TMP7]] to i64
// CHECK13-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM6]] // CHECK13-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM6]]
// CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group !6 // CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK13: omp.body.continue: // CHECK13: omp.body.continue:
// CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK13: omp.inner.for.inc: // CHECK13: omp.inner.for.inc:
// CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP8]], 1 // CHECK13-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK13-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK13-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK13: omp.inner.for.end: // CHECK13: omp.inner.for.end:
// CHECK13-NEXT: store i32 10, ptr [[I]], align 4 // CHECK13-NEXT: store i32 10, ptr [[I]], align 4
// CHECK13-NEXT: store i32 2, ptr [[J]], align 4 // CHECK13-NEXT: store i32 2, ptr [[J]], align 4
// CHECK13-NEXT: ret i32 0 // CHECK13-NEXT: ret i32 0
// //
// //
// CHECK15-LABEL: define {{[^@]+}}@main // CHECK15-LABEL: define {{[^@]+}}@main
▲ Show 20 Lines • Show All 59 Lines • ▼ Show 20 Lines
// CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK15-NEXT: [[CMP8:%.*]] = icmp slt i32 0, [[TMP10]] // CHECK15-NEXT: [[CMP8:%.*]] = icmp slt i32 0, [[TMP10]]
// CHECK15-NEXT: br i1 [[CMP8]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END]] // CHECK15-NEXT: br i1 [[CMP8]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END]]
// CHECK15: simd.if.then: // CHECK15: simd.if.then:
// CHECK15-NEXT: [[TMP11:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8 // CHECK15-NEXT: [[TMP11:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK15-NEXT: store i64 [[TMP11]], ptr [[DOTOMP_IV]], align 8 // CHECK15-NEXT: store i64 [[TMP11]], ptr [[DOTOMP_IV]], align 8
// CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK15: omp.inner.for.cond: // CHECK15: omp.inner.for.cond:
// CHECK15-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !3 // CHECK15-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3:![0-9]+]]
// CHECK15-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group !3 // CHECK15-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[CMP11:%.*]] = icmp sle i64 [[TMP12]], [[TMP13]] // CHECK15-NEXT: [[CMP11:%.*]] = icmp sle i64 [[TMP12]], [[TMP13]]
// CHECK15-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK15-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK15: omp.inner.for.body: // CHECK15: omp.inner.for.body:
// CHECK15-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !3 // CHECK15-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group !3 // CHECK15-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP15]], 0 // CHECK15-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP15]], 0
// CHECK15-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK15-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1
// CHECK15-NEXT: [[MUL14:%.*]] = mul nsw i32 1, [[DIV13]] // CHECK15-NEXT: [[MUL14:%.*]] = mul nsw i32 1, [[DIV13]]
// CHECK15-NEXT: [[CONV15:%.*]] = sext i32 [[MUL14]] to i64 // CHECK15-NEXT: [[CONV15:%.*]] = sext i32 [[MUL14]] to i64
// CHECK15-NEXT: [[DIV16:%.*]] = sdiv i64 [[TMP14]], [[CONV15]] // CHECK15-NEXT: [[DIV16:%.*]] = sdiv i64 [[TMP14]], [[CONV15]]
// CHECK15-NEXT: [[MUL17:%.*]] = mul nsw i64 [[DIV16]], 1 // CHECK15-NEXT: [[MUL17:%.*]] = mul nsw i64 [[DIV16]], 1
// CHECK15-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL17]] // CHECK15-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL17]]
// CHECK15-NEXT: [[CONV18:%.*]] = trunc i64 [[ADD]] to i32 // CHECK15-NEXT: [[CONV18:%.*]] = trunc i64 [[ADD]] to i32
// CHECK15-NEXT: store i32 [[CONV18]], ptr [[I9]], align 4, !llvm.access.group !3 // CHECK15-NEXT: store i32 [[CONV18]], ptr [[I9]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !3 // CHECK15-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !3 // CHECK15-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group !3 // CHECK15-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[SUB19:%.*]] = sub nsw i32 [[TMP18]], 0 // CHECK15-NEXT: [[SUB19:%.*]] = sub nsw i32 [[TMP18]], 0
// CHECK15-NEXT: [[DIV20:%.*]] = sdiv i32 [[SUB19]], 1 // CHECK15-NEXT: [[DIV20:%.*]] = sdiv i32 [[SUB19]], 1
// CHECK15-NEXT: [[MUL21:%.*]] = mul nsw i32 1, [[DIV20]] // CHECK15-NEXT: [[MUL21:%.*]] = mul nsw i32 1, [[DIV20]]
// CHECK15-NEXT: [[CONV22:%.*]] = sext i32 [[MUL21]] to i64 // CHECK15-NEXT: [[CONV22:%.*]] = sext i32 [[MUL21]] to i64
// CHECK15-NEXT: [[DIV23:%.*]] = sdiv i64 [[TMP17]], [[CONV22]] // CHECK15-NEXT: [[DIV23:%.*]] = sdiv i64 [[TMP17]], [[CONV22]]
// CHECK15-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group !3 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP19]], 0 // CHECK15-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP19]], 0
// CHECK15-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1 // CHECK15-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1
// CHECK15-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]] // CHECK15-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]]
// CHECK15-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64 // CHECK15-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64
// CHECK15-NEXT: [[MUL28:%.*]] = mul nsw i64 [[DIV23]], [[CONV27]] // CHECK15-NEXT: [[MUL28:%.*]] = mul nsw i64 [[DIV23]], [[CONV27]]
// CHECK15-NEXT: [[SUB29:%.*]] = sub nsw i64 [[TMP16]], [[MUL28]] // CHECK15-NEXT: [[SUB29:%.*]] = sub nsw i64 [[TMP16]], [[MUL28]]
// CHECK15-NEXT: [[MUL30:%.*]] = mul nsw i64 [[SUB29]], 1 // CHECK15-NEXT: [[MUL30:%.*]] = mul nsw i64 [[SUB29]], 1
// CHECK15-NEXT: [[ADD31:%.*]] = add nsw i64 0, [[MUL30]] // CHECK15-NEXT: [[ADD31:%.*]] = add nsw i64 0, [[MUL30]]
// CHECK15-NEXT: [[CONV32:%.*]] = trunc i64 [[ADD31]] to i32 // CHECK15-NEXT: [[CONV32:%.*]] = trunc i64 [[ADD31]] to i32
// CHECK15-NEXT: store i32 [[CONV32]], ptr [[J10]], align 4, !llvm.access.group !3 // CHECK15-NEXT: store i32 [[CONV32]], ptr [[J10]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[TMP20:%.*]] = load i32, ptr [[I9]], align 4, !llvm.access.group !3 // CHECK15-NEXT: [[TMP20:%.*]] = load i32, ptr [[I9]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[TMP21:%.*]] = mul nsw i32 [[TMP20]], [[TMP1]] // CHECK15-NEXT: [[TMP21:%.*]] = mul nsw i32 [[TMP20]], [[TMP1]]
// CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i32 [[TMP21]] // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i32 [[TMP21]]
// CHECK15-NEXT: [[TMP22:%.*]] = load i32, ptr [[J10]], align 4, !llvm.access.group !3 // CHECK15-NEXT: [[TMP22:%.*]] = load i32, ptr [[J10]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[ARRAYIDX33:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i32 [[TMP22]] // CHECK15-NEXT: [[ARRAYIDX33:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i32 [[TMP22]]
// CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX33]], align 4, !llvm.access.group !3 // CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX33]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK15: omp.body.continue: // CHECK15: omp.body.continue:
// CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK15: omp.inner.for.inc: // CHECK15: omp.inner.for.inc:
// CHECK15-NEXT: [[TMP23:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !3 // CHECK15-NEXT: [[TMP23:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[ADD34:%.*]] = add nsw i64 [[TMP23]], 1 // CHECK15-NEXT: [[ADD34:%.*]] = add nsw i64 [[TMP23]], 1
// CHECK15-NEXT: store i64 [[ADD34]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group !3 // CHECK15-NEXT: store i64 [[ADD34]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
// CHECK15: omp.inner.for.end: // CHECK15: omp.inner.for.end:
// CHECK15-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK15-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK15-NEXT: [[SUB35:%.*]] = sub nsw i32 [[TMP24]], 0 // CHECK15-NEXT: [[SUB35:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK15-NEXT: [[DIV36:%.*]] = sdiv i32 [[SUB35]], 1 // CHECK15-NEXT: [[DIV36:%.*]] = sdiv i32 [[SUB35]], 1
// CHECK15-NEXT: [[MUL37:%.*]] = mul nsw i32 [[DIV36]], 1 // CHECK15-NEXT: [[MUL37:%.*]] = mul nsw i32 [[DIV36]], 1
// CHECK15-NEXT: [[ADD38:%.*]] = add nsw i32 0, [[MUL37]] // CHECK15-NEXT: [[ADD38:%.*]] = add nsw i32 0, [[MUL37]]
// CHECK15-NEXT: store i32 [[ADD38]], ptr [[I9]], align 4 // CHECK15-NEXT: store i32 [[ADD38]], ptr [[I9]], align 4
Show All 28 Lines
// CHECK15-NEXT: [[J:%.*]] = alloca i32, align 4 // CHECK15-NEXT: [[J:%.*]] = alloca i32, align 4
// CHECK15-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK15-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK15-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK15-NEXT: store i32 19, ptr [[DOTOMP_UB]], align 4 // CHECK15-NEXT: store i32 19, ptr [[DOTOMP_UB]], align 4
// CHECK15-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK15-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK15-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 // CHECK15-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK15: omp.inner.for.cond: // CHECK15: omp.inner.for.cond:
// CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7:![0-9]+]]
// CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !7 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] // CHECK15-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK15-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK15: omp.inner.for.body: // CHECK15: omp.inner.for.body:
// CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 2 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 2
// CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK15-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !7 // CHECK15-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[DIV2:%.*]] = sdiv i32 [[TMP5]], 2 // CHECK15-NEXT: [[DIV2:%.*]] = sdiv i32 [[TMP5]], 2
// CHECK15-NEXT: [[MUL3:%.*]] = mul nsw i32 [[DIV2]], 2 // CHECK15-NEXT: [[MUL3:%.*]] = mul nsw i32 [[DIV2]], 2
// CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL3]] // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL3]]
// CHECK15-NEXT: [[MUL4:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK15-NEXT: [[MUL4:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK15-NEXT: [[ADD5:%.*]] = add nsw i32 0, [[MUL4]] // CHECK15-NEXT: [[ADD5:%.*]] = add nsw i32 0, [[MUL4]]
// CHECK15-NEXT: store i32 [[ADD5]], ptr [[J]], align 4, !llvm.access.group !7 // CHECK15-NEXT: store i32 [[ADD5]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !7 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[A]], i32 0, i32 [[TMP6]] // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[A]], i32 0, i32 [[TMP6]]
// CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group !7 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP7]] // CHECK15-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP7]]
// CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX6]], align 4, !llvm.access.group !7 // CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX6]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK15: omp.body.continue: // CHECK15: omp.body.continue:
// CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK15: omp.inner.for.inc: // CHECK15: omp.inner.for.inc:
// CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP8]], 1 // CHECK15-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK15-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK15-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]] // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK15: omp.inner.for.end: // CHECK15: omp.inner.for.end:
// CHECK15-NEXT: store i32 10, ptr [[I]], align 4 // CHECK15-NEXT: store i32 10, ptr [[I]], align 4
// CHECK15-NEXT: store i32 2, ptr [[J]], align 4 // CHECK15-NEXT: store i32 2, ptr [[J]], align 4
// CHECK15-NEXT: ret i32 0 // CHECK15-NEXT: ret i32 0
// //

clang/test/OpenMP/target_teams_distribute_parallel_for_simd_proc_bind_codegen.cpp

Show First 20 Lines • Show All 148 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 [ 999, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 999, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP4]], ptr [[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: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6:![0-9]+]]
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !6 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body: // CHECK1: omp.inner.for.body:
// CHECK1-NEXT: call void @__kmpc_push_proc_bind(ptr @[[GLOB3]], i32 [[TMP1]], i32 4), !llvm.access.group !6 // CHECK1-NEXT: call void @__kmpc_push_proc_bind(ptr @[[GLOB3]], i32 [[TMP1]], i32 4), !llvm.access.group [[ACC_GRP6]]
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !6 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !6 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 2, ptr @.omp_outlined..1, i64 [[TMP8]], i64 [[TMP10]]), !llvm.access.group !6 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 2, ptr @.omp_outlined..1, i64 [[TMP8]], i64 [[TMP10]]), !llvm.access.group [[ACC_GRP6]]
// 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: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group !6 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![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(ptr @[[GLOB1]], i32 [[TMP1]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK1-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
// CHECK1-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK1-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
▲ Show 20 Lines • Show All 45 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 [ 999, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 999, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP6]], ptr [[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: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !10 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP10:![0-9]+]]
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !10 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP10]]
// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// 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: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !10 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP10]]
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !10 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP10]]
// 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: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !10 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP10]]
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !10 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP10]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP11:![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(ptr @[[GLOB1]], i32 [[TMP3]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP3]])
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0 // CHECK1-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
// CHECK1-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK1-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
▲ Show 20 Lines • Show All 42 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 [ 999, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 999, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP4]], ptr [[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: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !15 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15:![0-9]+]]
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !15 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP15]]
// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body: // CHECK1: omp.inner.for.body:
// CHECK1-NEXT: call void @__kmpc_push_proc_bind(ptr @[[GLOB3]], i32 [[TMP1]], i32 3), !llvm.access.group !15 // CHECK1-NEXT: call void @__kmpc_push_proc_bind(ptr @[[GLOB3]], i32 [[TMP1]], i32 3), !llvm.access.group [[ACC_GRP15]]
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !15 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP15]]
// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !15 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP15]]
// CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 2, ptr @.omp_outlined..3, i64 [[TMP8]], i64 [[TMP10]]), !llvm.access.group !15 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 2, ptr @.omp_outlined..3, i64 [[TMP8]], i64 [[TMP10]]), !llvm.access.group [[ACC_GRP15]]
// 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: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !15 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15]]
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group !15 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP15]]
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !15 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![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(ptr @[[GLOB1]], i32 [[TMP1]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK1-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
// CHECK1-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK1-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
▲ Show 20 Lines • Show All 45 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 [ 999, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 999, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP6]], ptr [[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: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !18 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18:![0-9]+]]
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !18 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP18]]
// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// 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: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !18 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18]]
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !18 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP18]]
// 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: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !18 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18]]
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !18 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![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(ptr @[[GLOB1]], i32 [[TMP3]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP3]])
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0 // CHECK1-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
// CHECK1-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK1-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
▲ Show 20 Lines • Show All 75 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 [ 999, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 999, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP4]], ptr [[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: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !21 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP21:![0-9]+]]
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !21 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP21]]
// CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]] // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]
// CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1: omp.inner.for.body: // CHECK1: omp.inner.for.body:
// CHECK1-NEXT: call void @__kmpc_push_proc_bind(ptr @[[GLOB3]], i32 [[TMP1]], i32 2), !llvm.access.group !21 // CHECK1-NEXT: call void @__kmpc_push_proc_bind(ptr @[[GLOB3]], i32 [[TMP1]], i32 2), !llvm.access.group [[ACC_GRP21]]
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !21 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP21]]
// CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !21 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP21]]
// CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 2, ptr @.omp_outlined..5, i64 [[TMP8]], i64 [[TMP10]]), !llvm.access.group !21 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 2, ptr @.omp_outlined..5, i64 [[TMP8]], i64 [[TMP10]]), !llvm.access.group [[ACC_GRP21]]
// 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: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !21 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP21]]
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group !21 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP21]]
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !21 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP21]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP22:![0-9]+]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP22:![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(ptr @[[GLOB1]], i32 [[TMP1]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP1]])
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK1-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
// CHECK1-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK1-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
▲ Show 20 Lines • Show All 45 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 [ 999, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 999, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP6]], ptr [[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: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !24 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP24:![0-9]+]]
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !24 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP24]]
// CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// 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: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !24 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP24]]
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !24 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP24]]
// 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: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !24 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP24]]
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], 1
// CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !24 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP24]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP25:![0-9]+]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP25:![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(ptr @[[GLOB1]], i32 [[TMP3]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP3]])
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0 // CHECK1-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0
// CHECK1-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK1-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
Show All 27 Lines
// CHECK3-NEXT: [[I6:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[I6:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK3-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 999, ptr [[DOTOMP_UB]], align 4 // CHECK3-NEXT: store i32 999, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[TMP0]], ptr [[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: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2:![0-9]+]]
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !2 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK3-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] // CHECK3-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK3-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3: omp.inner.for.body: // CHECK3: omp.inner.for.body:
// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !2 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP2]]
// 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: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK3-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1 // CHECK3-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
// CHECK3-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK3-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK3: omp.inner.for.end: // CHECK3: omp.inner.for.end:
// CHECK3-NEXT: store i32 1000, ptr [[I]], align 4 // CHECK3-NEXT: store i32 1000, ptr [[I]], align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB3]], align 4 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB3]], align 4
// CHECK3-NEXT: store i32 999, ptr [[DOTOMP_UB4]], align 4 // CHECK3-NEXT: store i32 999, ptr [[DOTOMP_UB4]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB3]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB3]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV5]], align 4 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV5]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]]
// CHECK3: omp.inner.for.cond7: // CHECK3: omp.inner.for.cond7:
// CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV5]], align 4, !llvm.access.group !6 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV5]], align 4, !llvm.access.group [[ACC_GRP6:![0-9]+]]
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB4]], align 4, !llvm.access.group !6 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB4]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK3-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK3-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]] // CHECK3-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END15:%.*]]
// CHECK3: omp.inner.for.body9: // CHECK3: omp.inner.for.body9:
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV5]], align 4, !llvm.access.group !6 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV5]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1 // CHECK3-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK3-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]] // CHECK3-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]]
// CHECK3-NEXT: store i32 [[ADD11]], ptr [[I6]], align 4, !llvm.access.group !6 // CHECK3-NEXT: store i32 [[ADD11]], ptr [[I6]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]] // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]]
// CHECK3: omp.body.continue12: // CHECK3: omp.body.continue12:
// CHECK3-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]]
// CHECK3: omp.inner.for.inc13: // CHECK3: omp.inner.for.inc13:
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV5]], align 4, !llvm.access.group !6 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV5]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1 // CHECK3-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP9]], 1
// CHECK3-NEXT: store i32 [[ADD14]], ptr [[DOTOMP_IV5]], align 4, !llvm.access.group !6 // CHECK3-NEXT: store i32 [[ADD14]], ptr [[DOTOMP_IV5]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP7:![0-9]+]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK3: omp.inner.for.end15: // CHECK3: omp.inner.for.end15:
// CHECK3-NEXT: store i32 1000, ptr [[I6]], align 4 // CHECK3-NEXT: store i32 1000, ptr [[I6]], align 4
// CHECK3-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v() // CHECK3-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v()
// CHECK3-NEXT: ret i32 [[CALL]] // CHECK3-NEXT: ret i32 [[CALL]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@_Z5tmainIiET_v // CHECK3-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK3-SAME: () #[[ATTR1:[0-9]+]] comdat { // CHECK3-SAME: () #[[ATTR1:[0-9]+]] comdat {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 999, ptr [[DOTOMP_UB]], align 4 // CHECK3-NEXT: store i32 999, ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[TMP0]], ptr [[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: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !9 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9:![0-9]+]]
// CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !9 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK3-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] // CHECK3-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK3-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK3: omp.inner.for.body: // CHECK3: omp.inner.for.body:
// CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !9 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !9 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP9]]
// 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: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !9 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK3-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1 // CHECK3-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP4]], 1
// CHECK3-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !9 // CHECK3-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![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: store i32 1000, ptr [[I]], align 4 // CHECK3-NEXT: store i32 1000, ptr [[I]], align 4
// CHECK3-NEXT: ret i32 0 // CHECK3-NEXT: ret i32 0
// //

clang/test/OpenMP/target_teams_distribute_parallel_for_simd_reduction_codegen.cpp

Show First 20 Lines • Show All 95 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK1-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 8 // CHECK1-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 2, ptr [[TMP15]], align 8 // CHECK1-NEXT: store i64 2, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB4:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l66.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB4:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l66.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP15]], 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]+}}_main_l66(ptr @_ZZ4mainE5sivar) #[[ATTR2:[0-9]+]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l66(ptr @_ZZ4mainE5sivar) #[[ATTR2:[0-9]+]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v() // CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v()
// CHECK1-NEXT: ret i32 [[CALL]] // CHECK1-NEXT: ret i32 [[CALL]]
// //
// //
▲ Show 20 Lines • Show All 44 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 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !5 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5:![0-9]+]]
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !5 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP5]]
// 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, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !5 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !5 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB4]], i32 3, ptr @.omp_outlined..1, i64 [[TMP9]], i64 [[TMP11]], ptr [[SIVAR1]]), !llvm.access.group !5 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB4]], i32 3, ptr @.omp_outlined..1, i64 [[TMP9]], i64 [[TMP11]], ptr [[SIVAR1]]), !llvm.access.group [[ACC_GRP5]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !5 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group !5 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !5 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[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:
// CHECK1-NEXT: store i32 2, ptr [[I]], align 4 // CHECK1-NEXT: store i32 2, ptr [[I]], align 4
// CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK1: .omp.final.done: // CHECK1: .omp.final.done:
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[SIVAR1]], ptr [[TMP16]], align 8 // CHECK1-NEXT: store ptr [[SIVAR1]], ptr [[TMP16]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP19]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP20]], [[TMP21]] // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
// CHECK1-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK1-NEXT: [[TMP23:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP22]] monotonic, align 4 // CHECK1-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 41 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 [ 1, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP7]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !9 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9:![0-9]+]]
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !9 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP9]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !9 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !9 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !9 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[SIVAR2]], align 4, !llvm.access.group !9 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[SIVAR2]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP12]], [[TMP11]] // CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP12]], [[TMP11]]
// CHECK1-NEXT: store i32 [[ADD4]], ptr [[SIVAR2]], align 4, !llvm.access.group !9 // CHECK1-NEXT: store i32 [[ADD4]], ptr [[SIVAR2]], align 4, !llvm.access.group [[ACC_GRP9]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !9 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP13]], 1 // CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK1-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !9 // CHECK1-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![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(ptr @[[GLOB1]], i32 [[TMP4]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP4]])
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[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:
// CHECK1-NEXT: store i32 2, ptr [[I]], align 4 // CHECK1-NEXT: store i32 2, ptr [[I]], align 4
// CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK1: .omp.final.done: // CHECK1: .omp.final.done:
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[SIVAR2]], ptr [[TMP16]], align 8 // CHECK1-NEXT: store ptr [[SIVAR2]], ptr [[TMP16]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP19]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[SIVAR2]], align 4 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[SIVAR2]], align 4
// CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP20]], [[TMP21]] // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
// CHECK1-NEXT: store i32 [[ADD6]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD6]], ptr [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[SIVAR2]], align 4 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[SIVAR2]], align 4
// CHECK1-NEXT: [[TMP23:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP22]] monotonic, align 4 // CHECK1-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v // CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK1-SAME: () #[[ATTR5:[0-9]+]] comdat { // CHECK1-SAME: () #[[ATTR5:[0-9]+]] comdat {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[T_VAR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4 // CHECK1-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32 0, ptr [[T_VAR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[T_VAR]], align 4
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i64 8, i1 false) // CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i64 8, i1 false)
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[T_VAR]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[T_VAR]], ptr [[TMP1]], align 8 // CHECK1-NEXT: store ptr [[T_VAR]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[T_VAR]], ptr [[TMP3]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr null, ptr [[TMP5]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP9]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 8 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.7, ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.7, ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP13]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP15]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 2, ptr [[TMP16]], align 8 // CHECK1-NEXT: store i64 2, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB4]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB4]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP18]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP15]], 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]+}}__Z5tmainIiET_v_l32(ptr [[T_VAR]]) #[[ATTR2]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32(ptr [[T_VAR]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: ret i32 0 // CHECK1-NEXT: ret i32 0
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32
▲ Show 20 Lines • Show All 43 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 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !14 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14:![0-9]+]]
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !14 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP14]]
// 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, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !14 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP14]]
// CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !14 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP14]]
// CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB4]], i32 3, ptr @.omp_outlined..4, i64 [[TMP9]], i64 [[TMP11]], ptr [[T_VAR1]]), !llvm.access.group !14 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB4]], i32 3, ptr @.omp_outlined..4, i64 [[TMP9]], i64 [[TMP11]], ptr [[T_VAR1]]), !llvm.access.group [[ACC_GRP14]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !14 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group !14 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP14]]
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !14 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[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:
// CHECK1-NEXT: store i32 2, ptr [[I]], align 4 // CHECK1-NEXT: store i32 2, ptr [[I]], align 4
// CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK1: .omp.final.done: // CHECK1: .omp.final.done:
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[T_VAR1]], ptr [[TMP16]], align 8 // CHECK1-NEXT: store ptr [[T_VAR1]], ptr [[TMP16]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.6, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.6, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP19]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP20]], [[TMP21]] // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
// CHECK1-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK1-NEXT: [[TMP23:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP22]] monotonic, align 4 // CHECK1-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..4 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[T_VAR:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[T_VAR:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
▲ Show 20 Lines • Show All 41 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 [ 1, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP7]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !17 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17:![0-9]+]]
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !17 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP17]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !17 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !17 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !17 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[T_VAR2]], align 4, !llvm.access.group !17 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[T_VAR2]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP12]], [[TMP11]] // CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP12]], [[TMP11]]
// CHECK1-NEXT: store i32 [[ADD4]], ptr [[T_VAR2]], align 4, !llvm.access.group !17 // CHECK1-NEXT: store i32 [[ADD4]], ptr [[T_VAR2]], align 4, !llvm.access.group [[ACC_GRP17]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !17 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP13]], 1 // CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK1-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !17 // CHECK1-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP17]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP18:![0-9]+]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP18:![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(ptr @[[GLOB1]], i32 [[TMP4]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP4]])
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[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:
// CHECK1-NEXT: store i32 2, ptr [[I]], align 4 // CHECK1-NEXT: store i32 2, ptr [[I]], align 4
// CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK1: .omp.final.done: // CHECK1: .omp.final.done:
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[T_VAR2]], ptr [[TMP16]], align 8 // CHECK1-NEXT: store ptr [[T_VAR2]], ptr [[TMP16]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.5, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.5, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP19]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[T_VAR2]], align 4 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[T_VAR2]], align 4
// CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP20]], [[TMP21]] // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
// CHECK1-NEXT: store i32 [[ADD6]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD6]], ptr [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[T_VAR2]], align 4 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[T_VAR2]], align 4
// CHECK1-NEXT: [[TMP23:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP22]] monotonic, align 4 // CHECK1-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.5 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.5
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.6 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.6
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK1-SAME: () #[[ATTR7:[0-9]+]] { // CHECK1-SAME: () #[[ATTR7:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: call void @__tgt_register_requires(i64 1) // CHECK1-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@main // CHECK3-LABEL: define {{[^@]+}}@main
// CHECK3-SAME: () #[[ATTR0:[0-9]+]] { // CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK3-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 4 // CHECK3-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP2]], align 4 // CHECK3-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 2, ptr [[TMP15]], align 8 // CHECK3-NEXT: store i64 2, ptr [[TMP13]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB4:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l66.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB4:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l66.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP15]], 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]+}}_main_l66(ptr @_ZZ4mainE5sivar) #[[ATTR2:[0-9]+]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l66(ptr @_ZZ4mainE5sivar) #[[ATTR2:[0-9]+]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v() // CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v()
// CHECK3-NEXT: ret i32 [[CALL]] // CHECK3-NEXT: ret i32 [[CALL]]
// //
// //
▲ Show 20 Lines • Show All 44 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 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6:![0-9]+]]
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !6 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP6]]
// 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, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !6 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !6 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB4]], i32 3, ptr @.omp_outlined..1, i32 [[TMP8]], i32 [[TMP9]], ptr [[SIVAR1]]), !llvm.access.group !6 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB4]], i32 3, ptr @.omp_outlined..1, i32 [[TMP8]], i32 [[TMP9]], ptr [[SIVAR1]]), !llvm.access.group [[ACC_GRP6]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group !6 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[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:
// CHECK3-NEXT: store i32 2, ptr [[I]], align 4 // CHECK3-NEXT: store i32 2, ptr [[I]], align 4
// CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK3: .omp.final.done: // CHECK3: .omp.final.done:
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[SIVAR1]], ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr [[SIVAR1]], ptr [[TMP14]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3:[0-9]+]], i32 [[TMP2]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3:[0-9]+]], i32 [[TMP2]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK3-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK3-NEXT: ] // CHECK3-NEXT: ]
// CHECK3: .omp.reduction.case1: // CHECK3: .omp.reduction.case1:
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK3-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.case2: // CHECK3: .omp.reduction.case2:
// CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK3-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4 // CHECK3-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.default: // CHECK3: .omp.reduction.default:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..1 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR1]] { // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
Show All 39 Lines
// CHECK3-NEXT: br label [[COND_END]] // CHECK3-NEXT: br label [[COND_END]]
// CHECK3: cond.end: // CHECK3: cond.end:
// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[TMP7]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !10 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP10:![0-9]+]]
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !10 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP10]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !10 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP10]]
// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !10 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP10]]
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !10 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP10]]
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[SIVAR1]], align 4, !llvm.access.group !10 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[SIVAR1]], align 4, !llvm.access.group [[ACC_GRP10]]
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], [[TMP11]] // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], [[TMP11]]
// CHECK3-NEXT: store i32 [[ADD3]], ptr [[SIVAR1]], align 4, !llvm.access.group !10 // CHECK3-NEXT: store i32 [[ADD3]], ptr [[SIVAR1]], align 4, !llvm.access.group [[ACC_GRP10]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !10 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP10]]
// CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP13]], 1 // CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !10 // CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP10]]
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP11:![0-9]+]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP11:![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(ptr @[[GLOB1]], i32 [[TMP4]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP4]])
// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[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:
// CHECK3-NEXT: store i32 2, ptr [[I]], align 4 // CHECK3-NEXT: store i32 2, ptr [[I]], align 4
// CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK3: .omp.final.done: // CHECK3: .omp.final.done:
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[SIVAR1]], ptr [[TMP16]], align 4 // CHECK3-NEXT: store ptr [[SIVAR1]], ptr [[TMP16]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: switch i32 [[TMP19]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK3-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK3-NEXT: ] // CHECK3-NEXT: ]
// CHECK3: .omp.reduction.case1: // CHECK3: .omp.reduction.case1:
// CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP21:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP20]], [[TMP21]] // CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
// CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.case2: // CHECK3: .omp.reduction.case2:
// CHECK3-NEXT: [[TMP22:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK3-NEXT: [[TMP23:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP22]] monotonic, align 4 // CHECK3-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.default: // CHECK3: .omp.reduction.default:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { // CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2 // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@_Z5tmainIiET_v // CHECK3-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK3-SAME: () #[[ATTR5:[0-9]+]] comdat { // CHECK3-SAME: () #[[ATTR5:[0-9]+]] comdat {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[T_VAR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4 // CHECK3-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 0, ptr [[T_VAR]], align 4 // CHECK3-NEXT: store i32 0, ptr [[T_VAR]], align 4
// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i32 8, i1 false) // CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i32 8, i1 false)
// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR]], ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR]], ptr [[TMP1]], align 4 // CHECK3-NEXT: store ptr [[T_VAR]], ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR]], ptr [[TMP3]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP5]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, ptr [[TMP9]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 4 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes.7, ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes.7, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP13]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP15]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 2, ptr [[TMP16]], align 8 // CHECK3-NEXT: store i64 2, ptr [[TMP13]], align 8
// CHECK3-NEXT: [[TMP17:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB4]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB4]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP18]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP15]], 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]+}}__Z5tmainIiET_v_l32(ptr [[T_VAR]]) #[[ATTR2]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32(ptr [[T_VAR]]) #[[ATTR2]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: ret i32 0 // CHECK3-NEXT: ret i32 0
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32
▲ Show 20 Lines • Show All 43 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 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !15 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15:![0-9]+]]
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !15 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP15]]
// 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, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !15 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP15]]
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !15 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP15]]
// CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB4]], i32 3, ptr @.omp_outlined..4, i32 [[TMP8]], i32 [[TMP9]], ptr [[T_VAR1]]), !llvm.access.group !15 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB4]], i32 3, ptr @.omp_outlined..4, i32 [[TMP8]], i32 [[TMP9]], ptr [[T_VAR1]]), !llvm.access.group [[ACC_GRP15]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !15 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15]]
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group !15 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP15]]
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !15 // CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15]]
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[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:
// CHECK3-NEXT: store i32 2, ptr [[I]], align 4 // CHECK3-NEXT: store i32 2, ptr [[I]], align 4
// CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK3: .omp.final.done: // CHECK3: .omp.final.done:
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR1]], ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr [[T_VAR1]], ptr [[TMP14]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.6, ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.6, ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK3-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK3-NEXT: ] // CHECK3-NEXT: ]
// CHECK3: .omp.reduction.case1: // CHECK3: .omp.reduction.case1:
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK3-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.case2: // CHECK3: .omp.reduction.case2:
// CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK3-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4 // CHECK3-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.default: // CHECK3: .omp.reduction.default:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..4 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..4
// CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[T_VAR:%.*]]) #[[ATTR1]] { // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[T_VAR:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
Show All 39 Lines
// CHECK3-NEXT: br label [[COND_END]] // CHECK3-NEXT: br label [[COND_END]]
// CHECK3: cond.end: // CHECK3: cond.end:
// CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[TMP7]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !18 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18:![0-9]+]]
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !18 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP18]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !18 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18]]
// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !18 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP18]]
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !18 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP18]]
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group !18 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP18]]
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], [[TMP11]] // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], [[TMP11]]
// CHECK3-NEXT: store i32 [[ADD3]], ptr [[T_VAR1]], align 4, !llvm.access.group !18 // CHECK3-NEXT: store i32 [[ADD3]], ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP18]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !18 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18]]
// CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP13]], 1 // CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !18 // CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP18]]
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![0-9]+]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP19:![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(ptr @[[GLOB1]], i32 [[TMP4]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP4]])
// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[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:
// CHECK3-NEXT: store i32 2, ptr [[I]], align 4 // CHECK3-NEXT: store i32 2, ptr [[I]], align 4
// CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK3: .omp.final.done: // CHECK3: .omp.final.done:
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR1]], ptr [[TMP16]], align 4 // CHECK3-NEXT: store ptr [[T_VAR1]], ptr [[TMP16]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.5, ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.5, ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: switch i32 [[TMP19]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK3-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK3-NEXT: ] // CHECK3-NEXT: ]
// CHECK3: .omp.reduction.case1: // CHECK3: .omp.reduction.case1:
// CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP21:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP20]], [[TMP21]] // CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
// CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.case2: // CHECK3: .omp.reduction.case2:
// CHECK3-NEXT: [[TMP22:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK3-NEXT: [[TMP23:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP22]] monotonic, align 4 // CHECK3-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.default: // CHECK3: .omp.reduction.default:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.5 // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.5
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.6 // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.6
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK3-SAME: () #[[ATTR7:[0-9]+]] { // CHECK3-SAME: () #[[ATTR7:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: call void @__tgt_register_requires(i64 1) // CHECK3-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
▲ Show 20 Lines • Show All 56 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: br label [[COND_END]] // CHECK5-NEXT: br label [[COND_END]]
// CHECK5: cond.end: // CHECK5: cond.end:
// CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK5-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 // CHECK5-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond: // CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4:![0-9]+]]
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !4 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] // CHECK5-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]]
// CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body: // CHECK5: omp.inner.for.body:
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group !4 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK5-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 // CHECK5-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64
// CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group !4 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK5-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 // CHECK5-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64
// CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB4]], i32 3, ptr @.omp_outlined..1, i64 [[TMP9]], i64 [[TMP11]], ptr [[SIVAR1]]), !llvm.access.group !4 // CHECK5-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB4]], i32 3, ptr @.omp_outlined..1, i64 [[TMP9]], i64 [[TMP11]], ptr [[SIVAR1]]), !llvm.access.group [[ACC_GRP4]]
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc: // CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group !4 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK5-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]
// CHECK5: omp.inner.for.end: // CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit: // CHECK5: omp.loop.exit:
// CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0 // CHECK5-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK5-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK5-NEXT: br i1 [[TMP15]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK5: .omp.final.then: // CHECK5: .omp.final.then:
// CHECK5-NEXT: store i32 2, ptr [[I]], align 4 // CHECK5-NEXT: store i32 2, ptr [[I]], align 4
// CHECK5-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK5-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK5: .omp.final.done: // CHECK5: .omp.final.done:
// CHECK5-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK5-NEXT: [[TMP16:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK5-NEXT: store ptr [[SIVAR1]], ptr [[TMP16]], align 8 // CHECK5-NEXT: store ptr [[SIVAR1]], ptr [[TMP16]], align 8
// CHECK5-NEXT: [[TMP19:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: switch i32 [[TMP19]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK5-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK5-NEXT: ] // CHECK5-NEXT: ]
// CHECK5: .omp.reduction.case1: // CHECK5: .omp.reduction.case1:
// CHECK5-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK5-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK5-NEXT: [[TMP21:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK5-NEXT: [[TMP19:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP20]], [[TMP21]] // CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]]
// CHECK5-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4 // CHECK5-NEXT: store i32 [[ADD3]], ptr [[TMP0]], align 4
// CHECK5-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.case2: // CHECK5: .omp.reduction.case2:
// CHECK5-NEXT: [[TMP22:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK5-NEXT: [[TMP20:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK5-NEXT: [[TMP23:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP22]] monotonic, align 4 // CHECK5-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.default: // CHECK5: .omp.reduction.default:
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..1 // CHECK5-LABEL: define {{[^@]+}}@.omp_outlined..1
// CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR2]] { // CHECK5-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[SIVAR:%.*]]) #[[ATTR2]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
▲ Show 20 Lines • Show All 42 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: br label [[COND_END]] // CHECK5-NEXT: br label [[COND_END]]
// CHECK5: cond.end: // CHECK5: cond.end:
// CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] // CHECK5-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ]
// CHECK5-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 // CHECK5-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond: // CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !8 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8:![0-9]+]]
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !8 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP8]]
// CHECK5-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] // CHECK5-NEXT: [[CMP3:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]]
// CHECK5-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5-NEXT: br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body: // CHECK5: omp.inner.for.body:
// CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !8 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8]]
// CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !8 // CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP8]]
// CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !8 // CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP8]]
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[SIVAR2]], align 4, !llvm.access.group !8 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[SIVAR2]], align 4, !llvm.access.group [[ACC_GRP8]]
// CHECK5-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP12]], [[TMP11]] // CHECK5-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP12]], [[TMP11]]
// CHECK5-NEXT: store i32 [[ADD4]], ptr [[SIVAR2]], align 4, !llvm.access.group !8 // CHECK5-NEXT: store i32 [[ADD4]], ptr [[SIVAR2]], align 4, !llvm.access.group [[ACC_GRP8]]
// CHECK5-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[REF_TMP]], i32 0, i32 0 // CHECK5-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[REF_TMP]], i32 0, i32 0
// CHECK5-NEXT: store ptr [[SIVAR2]], ptr [[TMP13]], align 8, !llvm.access.group !8 // CHECK5-NEXT: store ptr [[SIVAR2]], ptr [[TMP13]], align 8, !llvm.access.group [[ACC_GRP8]]
// CHECK5-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(ptr noundef nonnull align 8 dereferenceable(8) [[REF_TMP]]), !llvm.access.group !8 // CHECK5-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(ptr noundef nonnull align 8 dereferenceable(8) [[REF_TMP]]), !llvm.access.group [[ACC_GRP8]]
// CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK5: omp.body.continue: // CHECK5: omp.body.continue:
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc: // CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !8 // CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8]]
// CHECK5-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP14]], 1 // CHECK5-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP14]], 1
// CHECK5-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !8 // CHECK5-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8]]
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]]
// CHECK5: omp.inner.for.end: // CHECK5: omp.inner.for.end:
// CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]] // CHECK5-NEXT: br label [[OMP_LOOP_EXIT:%.*]]
// CHECK5: omp.loop.exit: // CHECK5: omp.loop.exit:
// CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP4]]) // CHECK5-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP4]])
// CHECK5-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK5-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK5-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0 // CHECK5-NEXT: [[TMP16:%.*]] = icmp ne i32 [[TMP15]], 0
// CHECK5-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK5-NEXT: br i1 [[TMP16]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK5: .omp.final.then: // CHECK5: .omp.final.then:
// CHECK5-NEXT: store i32 2, ptr [[I]], align 4 // CHECK5-NEXT: store i32 2, ptr [[I]], align 4
// CHECK5-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK5-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK5: .omp.final.done: // CHECK5: .omp.final.done:
// CHECK5-NEXT: [[TMP17:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK5-NEXT: [[TMP17:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK5-NEXT: store ptr [[SIVAR2]], ptr [[TMP17]], align 8 // CHECK5-NEXT: store ptr [[SIVAR2]], ptr [[TMP17]], align 8
// CHECK5-NEXT: [[TMP20:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: [[TMP18:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: switch i32 [[TMP20]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK5-NEXT: switch i32 [[TMP18]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK5-NEXT: ] // CHECK5-NEXT: ]
// CHECK5: .omp.reduction.case1: // CHECK5: .omp.reduction.case1:
// CHECK5-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK5-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK5-NEXT: [[TMP22:%.*]] = load i32, ptr [[SIVAR2]], align 4 // CHECK5-NEXT: [[TMP20:%.*]] = load i32, ptr [[SIVAR2]], align 4
// CHECK5-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] // CHECK5-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP19]], [[TMP20]]
// CHECK5-NEXT: store i32 [[ADD6]], ptr [[TMP0]], align 4 // CHECK5-NEXT: store i32 [[ADD6]], ptr [[TMP0]], align 4
// CHECK5-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: call void @__kmpc_end_reduce_nowait(ptr @[[GLOB3]], i32 [[TMP4]], ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.case2: // CHECK5: .omp.reduction.case2:
// CHECK5-NEXT: [[TMP23:%.*]] = load i32, ptr [[SIVAR2]], align 4 // CHECK5-NEXT: [[TMP21:%.*]] = load i32, ptr [[SIVAR2]], align 4
// CHECK5-NEXT: [[TMP24:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP23]] monotonic, align 4 // CHECK5-NEXT: [[TMP22:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP21]] monotonic, align 4
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.default: // CHECK5: .omp.reduction.default:
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4:[0-9]+]] { // CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK5-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK5-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK5-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK5-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2 // CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2
// CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4]] { // CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK5-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK5-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK5-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK5-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK5-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK5-SAME: () #[[ATTR6:[0-9]+]] { // CHECK5-SAME: () #[[ATTR6:[0-9]+]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: call void @__tgt_register_requires(i64 1) // CHECK5-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
Show All 12 Lines
// CHECK7-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK7-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK7-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4 // CHECK7-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK7-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK7-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK7-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 // CHECK7-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK7-NEXT: store i32 0, ptr [[SIVAR]], align 4 // CHECK7-NEXT: store i32 0, ptr [[SIVAR]], align 4
// CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK7: omp.inner.for.cond: // CHECK7: omp.inner.for.cond:
// CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2:![0-9]+]]
// CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !2 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] // CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK7: omp.inner.for.body: // CHECK7: omp.inner.for.body:
// CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 // CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !2 // CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !2 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[SIVAR]], align 4, !llvm.access.group !2 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[SIVAR]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK7-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], [[TMP4]] // CHECK7-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], [[TMP4]]
// CHECK7-NEXT: store i32 [[ADD1]], ptr [[SIVAR]], align 4, !llvm.access.group !2 // CHECK7-NEXT: store i32 [[ADD1]], ptr [[SIVAR]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK7: omp.body.continue: // CHECK7: omp.body.continue:
// CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK7: omp.inner.for.inc: // CHECK7: omp.inner.for.inc:
// CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK7-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], 1 // CHECK7-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK7-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK7-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK7: omp.inner.for.end: // CHECK7: omp.inner.for.end:
// CHECK7-NEXT: store i32 2, ptr [[I]], align 4 // CHECK7-NEXT: store i32 2, ptr [[I]], align 4
// CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4
// CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[SIVAR]], align 4 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[SIVAR]], align 4
// CHECK7-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP7]], [[TMP8]] // CHECK7-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
// CHECK7-NEXT: store i32 [[ADD3]], ptr @_ZZ4mainE5sivar, align 4 // CHECK7-NEXT: store i32 [[ADD3]], ptr @_ZZ4mainE5sivar, align 4
// CHECK7-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v() // CHECK7-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v()
Show All 10 Lines
// CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[T_VAR1:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[T_VAR1:%.*]] = alloca i32, align 4
// CHECK7-NEXT: store i32 0, ptr [[T_VAR]], align 4 // CHECK7-NEXT: store i32 0, ptr [[T_VAR]], align 4
// CHECK7-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i64 8, i1 false) // CHECK7-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i64 8, i1 false)
// CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK7-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4 // CHECK7-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK7-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK7-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_IV]], align 4 // CHECK7-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK7-NEXT: store i32 0, ptr [[T_VAR1]], align 4 // CHECK7-NEXT: store i32 0, ptr [[T_VAR1]], align 4
// CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK7: omp.inner.for.cond: // CHECK7: omp.inner.for.cond:
// CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6:![0-9]+]]
// CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !6 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP2]], [[TMP3]] // CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK7: omp.inner.for.body: // CHECK7: omp.inner.for.body:
// CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP4]], 1 // CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !6 // CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !6 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group !6 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK7-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], [[TMP5]] // CHECK7-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP5]], [[TMP4]]
// CHECK7-NEXT: store i32 [[ADD2]], ptr [[T_VAR1]], align 4, !llvm.access.group !6 // CHECK7-NEXT: store i32 [[ADD2]], ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK7: omp.body.continue: // CHECK7: omp.body.continue:
// CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK7: omp.inner.for.inc: // CHECK7: omp.inner.for.inc:
// CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK7-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP7]], 1 // CHECK7-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK7-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK7-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK7: omp.inner.for.end: // CHECK7: omp.inner.for.end:
// CHECK7-NEXT: store i32 2, ptr [[I]], align 4 // CHECK7-NEXT: store i32 2, ptr [[I]], align 4
// CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[T_VAR]], align 4 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[T_VAR]], align 4
// CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK7-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP8]], [[TMP9]] // CHECK7-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
// CHECK7-NEXT: store i32 [[ADD4]], ptr [[T_VAR]], align 4 // CHECK7-NEXT: store i32 [[ADD4]], ptr [[T_VAR]], align 4
// CHECK7-NEXT: ret i32 0 // CHECK7-NEXT: ret i32 0
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@main // CHECK9-LABEL: define {{[^@]+}}@main
// CHECK9-SAME: () #[[ATTR0:[0-9]+]] { // CHECK9-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[RETVAL:%.*]] = 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
// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[SIVAR:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[SIVAR:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT: store i32 0, ptr [[SIVAR]], align 4 // CHECK9-NEXT: store i32 0, ptr [[SIVAR]], 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: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3:![0-9]+]]
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !3 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK9-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] // CHECK9-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK9-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body: // CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !3 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !3 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[SIVAR]], align 4, !llvm.access.group !3 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[SIVAR]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK9-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], [[TMP4]] // CHECK9-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], [[TMP4]]
// CHECK9-NEXT: store i32 [[ADD1]], ptr [[SIVAR]], align 4, !llvm.access.group !3 // CHECK9-NEXT: store i32 [[ADD1]], ptr [[SIVAR]], align 4, !llvm.access.group [[ACC_GRP3]]
// 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: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK9-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], 1 // CHECK9-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK9-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK9-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: store i32 2, ptr [[I]], align 4 // CHECK9-NEXT: store i32 2, ptr [[I]], align 4
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[SIVAR]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[SIVAR]], align 4
// CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP7]], [[TMP8]] // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
// CHECK9-NEXT: store i32 [[ADD3]], ptr @_ZZ4mainE5sivar, align 4 // CHECK9-NEXT: store i32 [[ADD3]], ptr @_ZZ4mainE5sivar, align 4
// CHECK9-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v() // CHECK9-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v()
Show All 10 Lines
// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[T_VAR1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[T_VAR1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 0, ptr [[T_VAR]], align 4 // CHECK9-NEXT: store i32 0, ptr [[T_VAR]], align 4
// CHECK9-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i32 8, i1 false) // CHECK9-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i32 8, i1 false)
// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK9-NEXT: store i32 0, ptr [[T_VAR1]], align 4 // CHECK9-NEXT: store i32 0, ptr [[T_VAR1]], 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: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7:![0-9]+]]
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !7 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK9-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP2]], [[TMP3]] // CHECK9-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK9-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK9: omp.inner.for.body: // CHECK9: omp.inner.for.body:
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP4]], 1 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !7 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !7 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group !7 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK9-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], [[TMP5]] // CHECK9-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP5]], [[TMP4]]
// CHECK9-NEXT: store i32 [[ADD2]], ptr [[T_VAR1]], align 4, !llvm.access.group !7 // CHECK9-NEXT: store i32 [[ADD2]], ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP7]]
// 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: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP7]], 1 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK9-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK9-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK9: omp.inner.for.end: // CHECK9: omp.inner.for.end:
// CHECK9-NEXT: store i32 2, ptr [[I]], align 4 // CHECK9-NEXT: store i32 2, ptr [[I]], align 4
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[T_VAR]], align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[T_VAR]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK9-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP8]], [[TMP9]] // CHECK9-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
// CHECK9-NEXT: store i32 [[ADD4]], ptr [[T_VAR]], align 4 // CHECK9-NEXT: store i32 [[ADD4]], ptr [[T_VAR]], align 4
// CHECK9-NEXT: ret i32 0 // CHECK9-NEXT: ret i32 0
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@main // CHECK11-LABEL: define {{[^@]+}}@main
// CHECK11-SAME: () #[[ATTR0:[0-9]+]] { // CHECK11-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK11-NEXT: entry: // CHECK11-NEXT: entry:
// CHECK11-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK11-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK11-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 1 // CHECK11-NEXT: [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 1
// CHECK11-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK11-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK11-NEXT: call void @"_ZZ4mainENK3$_0clEv"(ptr noundef nonnull align 1 dereferenceable(1) [[REF_TMP]]) // CHECK11-NEXT: call void @"_ZZ4mainENK3$_0clEv"(ptr noundef nonnull align 1 dereferenceable(1) [[REF_TMP]])
// CHECK11-NEXT: ret i32 0 // CHECK11-NEXT: ret i32 0
// //

clang/test/OpenMP/target_teams_distribute_reduction_codegen.cpp

Show First 20 Lines • Show All 83 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK1-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 8 // CHECK1-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 2, ptr [[TMP15]], align 8 // CHECK1-NEXT: store i64 2, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l60.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l60.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP15]], 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]+}}_main_l60(ptr @_ZZ4mainE5sivar) #[[ATTR2:[0-9]+]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l60(ptr @_ZZ4mainE5sivar) #[[ATTR2:[0-9]+]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v() // CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v()
// CHECK1-NEXT: ret i32 [[CALL]] // CHECK1-NEXT: ret i32 [[CALL]]
// //
// //
▲ Show 20 Lines • Show All 71 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[SIVAR1]], ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr [[SIVAR1]], ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP13:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP13]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP16]], [[TMP17]] // CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK1-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4 // CHECK1-NEXT: [[TMP17:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP16]] monotonic, align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v // CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK1-SAME: () #[[ATTR5:[0-9]+]] comdat { // CHECK1-SAME: () #[[ATTR5:[0-9]+]] comdat {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[T_VAR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4 // CHECK1-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32 0, ptr [[T_VAR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[T_VAR]], align 4
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i64 8, i1 false) // CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i64 8, i1 false)
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[T_VAR]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[T_VAR]], ptr [[TMP1]], align 8 // CHECK1-NEXT: store ptr [[T_VAR]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[T_VAR]], ptr [[TMP3]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr null, ptr [[TMP5]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP9]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 8 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.3, ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.3, ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP13]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP15]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 2, ptr [[TMP16]], align 8 // CHECK1-NEXT: store i64 2, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP18]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP15]], 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]+}}__Z5tmainIiET_v_l32(ptr [[T_VAR]]) #[[ATTR2]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32(ptr [[T_VAR]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: ret i32 0 // CHECK1-NEXT: ret i32 0
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32
▲ Show 20 Lines • Show All 70 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[T_VAR1]], ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr [[T_VAR1]], ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP13:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP13]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP16]], [[TMP17]] // CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK1-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4 // CHECK1-NEXT: [[TMP17:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP16]] monotonic, align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK1-SAME: () #[[ATTR7:[0-9]+]] { // CHECK1-SAME: () #[[ATTR7:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: call void @__tgt_register_requires(i64 1) // CHECK1-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@main // CHECK3-LABEL: define {{[^@]+}}@main
// CHECK3-SAME: () #[[ATTR0:[0-9]+]] { // CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK3-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 4 // CHECK3-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP2]], align 4 // CHECK3-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 2, ptr [[TMP15]], align 8 // CHECK3-NEXT: store i64 2, ptr [[TMP13]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l60.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l60.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP15]], 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]+}}_main_l60(ptr @_ZZ4mainE5sivar) #[[ATTR2:[0-9]+]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l60(ptr @_ZZ4mainE5sivar) #[[ATTR2:[0-9]+]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v() // CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v()
// CHECK3-NEXT: ret i32 [[CALL]] // CHECK3-NEXT: ret i32 [[CALL]]
// //
// //
▲ Show 20 Lines • Show All 71 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[SIVAR1]], ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr [[SIVAR1]], ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: [[TMP13:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK3-NEXT: switch i32 [[TMP13]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK3-NEXT: ] // CHECK3-NEXT: ]
// CHECK3: .omp.reduction.case1: // CHECK3: .omp.reduction.case1:
// CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP16]], [[TMP17]] // CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.case2: // CHECK3: .omp.reduction.case2:
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4 // CHECK3-NEXT: [[TMP17:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP16]] monotonic, align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.default: // CHECK3: .omp.reduction.default:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { // CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@_Z5tmainIiET_v // CHECK3-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK3-SAME: () #[[ATTR5:[0-9]+]] comdat { // CHECK3-SAME: () #[[ATTR5:[0-9]+]] comdat {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[T_VAR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4 // CHECK3-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 0, ptr [[T_VAR]], align 4 // CHECK3-NEXT: store i32 0, ptr [[T_VAR]], align 4
// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i32 8, i1 false) // CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i32 8, i1 false)
// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR]], ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR]], ptr [[TMP1]], align 4 // CHECK3-NEXT: store ptr [[T_VAR]], ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR]], ptr [[TMP3]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP5]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, ptr [[TMP9]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 4 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes.3, ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes.3, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP13]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP15]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 2, ptr [[TMP16]], align 8 // CHECK3-NEXT: store i64 2, ptr [[TMP13]], align 8
// CHECK3-NEXT: [[TMP17:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP18]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP15]], 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]+}}__Z5tmainIiET_v_l32(ptr [[T_VAR]]) #[[ATTR2]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32(ptr [[T_VAR]]) #[[ATTR2]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: ret i32 0 // CHECK3-NEXT: ret i32 0
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32
▲ Show 20 Lines • Show All 70 Lines • ▼ Show 20 Lines
// CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]]
// 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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR1]], ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr [[T_VAR1]], ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP2]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: [[TMP13:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP2]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK3-NEXT: switch i32 [[TMP13]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK3-NEXT: ] // CHECK3-NEXT: ]
// CHECK3: .omp.reduction.case1: // CHECK3: .omp.reduction.case1:
// CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP16]], [[TMP17]] // CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.case2: // CHECK3: .omp.reduction.case2:
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4 // CHECK3-NEXT: [[TMP17:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP16]] monotonic, align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.default: // CHECK3: .omp.reduction.default:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2 // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK3-SAME: () #[[ATTR7:[0-9]+]] { // CHECK3-SAME: () #[[ATTR7:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: call void @__tgt_register_requires(i64 1) // CHECK3-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
▲ Show 20 Lines • Show All 87 Lines • ▼ Show 20 Lines
// CHECK9-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[ADD4]], ptr [[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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK9-NEXT: store ptr [[SIVAR1]], ptr [[TMP13]], align 8 // CHECK9-NEXT: store ptr [[SIVAR1]], ptr [[TMP13]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var) // CHECK9-NEXT: [[TMP14:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK9-NEXT: switch i32 [[TMP16]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK9-NEXT: switch i32 [[TMP14]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK9-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK9-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK9-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK9-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK9-NEXT: ] // CHECK9-NEXT: ]
// CHECK9: .omp.reduction.case1: // CHECK9: .omp.reduction.case1:
// CHECK9-NEXT: [[TMP17:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK9-NEXT: [[TMP18:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK9-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP17]], [[TMP18]] // CHECK9-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP15]], [[TMP16]]
// CHECK9-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4 // CHECK9-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK9-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK9-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK9-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK9-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK9: .omp.reduction.case2: // CHECK9: .omp.reduction.case2:
// CHECK9-NEXT: [[TMP19:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK9-NEXT: [[TMP20:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP19]] monotonic, align 4 // CHECK9-NEXT: [[TMP18:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP17]] monotonic, align 4
// CHECK9-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK9-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK9-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK9-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK9: .omp.reduction.default: // CHECK9: .omp.reduction.default:
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK9-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK9-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4:[0-9]+]] { // CHECK9-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK9-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK9-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK9-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK9-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK9-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK9-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK9-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK9-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK9-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK9-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK9-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK9-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK9-SAME: () #[[ATTR6:[0-9]+]] { // CHECK9-SAME: () #[[ATTR6:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: call void @__tgt_register_requires(i64 1) // CHECK9-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //

clang/test/OpenMP/target_teams_distribute_simd_collapse_codegen.cpp

Show First 20 Lines • Show All 110 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 8 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[A]], ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr [[A]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 56088, ptr [[TMP15]], align 8 // CHECK1-NEXT: store i64 56088, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP15]], 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]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR3:[0-9]+]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR3:[0-9]+]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 // CHECK1-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i64 0, i64 0 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i64 0, i64 0
// CHECK1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 0 // CHECK1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 0
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4
// CHECK1-NEXT: ret i32 [[TMP18]] // CHECK1-NEXT: ret i32 [[TMP16]]
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28
// CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
Show All 37 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]], ptr [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4:![0-9]+]]
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP4]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// 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]], ptr [[I]], align 4, !llvm.access.group !4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// 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]], ptr [[J]], align 4, !llvm.access.group !4 // CHECK1-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP4]]
// 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]], ptr [[A]], i64 0, i64 [[IDXPROM]] // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK1-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP12]] to i64 // CHECK1-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP12]] to i64
// CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]] // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]]
// CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX8]], align 4, !llvm.access.group !4 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX8]], align 4, !llvm.access.group [[ACC_GRP4]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK1-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP13]], 1 // CHECK1-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK1-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK1-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[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:%.*]]
Show All 29 Lines
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[_TMP2:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[_TMP2:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[A]], ptr [[TMP2]], align 4 // CHECK3-NEXT: store ptr [[A]], ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 56088, ptr [[TMP15]], align 8 // CHECK3-NEXT: store i64 56088, ptr [[TMP13]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP15]], 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]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR3:[0-9]+]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR3:[0-9]+]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 // CHECK3-NEXT: [[A3:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i32 0, i32 0 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A3]], i32 0, i32 0
// CHECK3-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 0 // CHECK3-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 0
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4
// CHECK3-NEXT: ret i32 [[TMP18]] // CHECK3-NEXT: ret i32 [[TMP16]]
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28
// CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { // CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
Show All 37 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]], ptr [[DOTOMP_UB]], align 4 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !5 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5:![0-9]+]]
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !5 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP5]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !5 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
// 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]], ptr [[I]], align 4, !llvm.access.group !5 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !5 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !5 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
// 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]], ptr [[J]], align 4, !llvm.access.group !5 // CHECK3-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !5 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i32 0, i32 [[TMP11]] // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i32 0, i32 [[TMP11]]
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group !5 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK3-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP12]] // CHECK3-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP12]]
// CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group !5 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group [[ACC_GRP5]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !5 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP13]], 1 // CHECK3-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK3-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !5 // CHECK3-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[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:%.*]]
Show All 34 Lines
// CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 // CHECK5-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK5-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 // CHECK5-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 56087, ptr [[DOTOMP_UB]], align 4 // CHECK5-NEXT: store i32 56087, ptr [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond: // CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2:![0-9]+]]
// CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] // CHECK5-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK5-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body: // CHECK5: omp.inner.for.body:
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 456 // CHECK5-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 456
// CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !2 // CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP5]], 456 // CHECK5-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP5]], 456
// CHECK5-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456 // CHECK5-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456
// CHECK5-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL4]] // CHECK5-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL4]]
// CHECK5-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK5-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK5-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]] // CHECK5-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK5-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group !2 // CHECK5-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 // CHECK5-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP6]] to i64 // CHECK5-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP6]] to i64
// CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i64 0, i64 [[IDXPROM]] // CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i64 0, i64 [[IDXPROM]]
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP7]] to i64 // CHECK5-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP7]] to i64
// CHECK5-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]] // CHECK5-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]]
// CHECK5-NEXT: store i32 0, ptr [[ARRAYIDX8]], align 4, !llvm.access.group !2 // CHECK5-NEXT: store i32 0, ptr [[ARRAYIDX8]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK5: omp.body.continue: // CHECK5: omp.body.continue:
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc: // CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP8]], 1 // CHECK5-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK5-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK5-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK5: omp.inner.for.end: // CHECK5: omp.inner.for.end:
// CHECK5-NEXT: store i32 123, ptr [[I]], align 4 // CHECK5-NEXT: store i32 123, ptr [[I]], align 4
// CHECK5-NEXT: store i32 456, ptr [[J]], align 4 // CHECK5-NEXT: store i32 456, ptr [[J]], align 4
// CHECK5-NEXT: [[A10:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 // CHECK5-NEXT: [[A10:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK5-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A10]], i64 0, i64 0 // CHECK5-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A10]], i64 0, i64 0
// CHECK5-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX11]], i64 0, i64 0 // CHECK5-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX11]], i64 0, i64 0
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX12]], align 4 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX12]], align 4
Show All 22 Lines
// CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 // CHECK7-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4
// CHECK7-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 // CHECK7-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4
// CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK7-NEXT: store i32 56087, ptr [[DOTOMP_UB]], align 4 // CHECK7-NEXT: store i32 56087, ptr [[DOTOMP_UB]], align 4
// CHECK7-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK7-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK7-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 // CHECK7-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK7: omp.inner.for.cond: // CHECK7: omp.inner.for.cond:
// CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3:![0-9]+]]
// CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] // CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK7: omp.inner.for.body: // CHECK7: omp.inner.for.body:
// CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 456 // CHECK7-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 456
// CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !3 // CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP5]], 456 // CHECK7-NEXT: [[DIV3:%.*]] = sdiv i32 [[TMP5]], 456
// CHECK7-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456 // CHECK7-NEXT: [[MUL4:%.*]] = mul nsw i32 [[DIV3]], 456
// CHECK7-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL4]] // CHECK7-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL4]]
// CHECK7-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK7-NEXT: [[MUL5:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK7-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]] // CHECK7-NEXT: [[ADD6:%.*]] = add nsw i32 0, [[MUL5]]
// CHECK7-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group !3 // CHECK7-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 // CHECK7-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i32 0, i32 [[TMP6]] // CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A]], i32 0, i32 [[TMP6]]
// CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP7]] // CHECK7-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP7]]
// CHECK7-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group !3 // CHECK7-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK7: omp.body.continue: // CHECK7: omp.body.continue:
// CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK7: omp.inner.for.inc: // CHECK7: omp.inner.for.inc:
// CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP8]], 1 // CHECK7-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK7-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK7-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
// CHECK7: omp.inner.for.end: // CHECK7: omp.inner.for.end:
// CHECK7-NEXT: store i32 123, ptr [[I]], align 4 // CHECK7-NEXT: store i32 123, ptr [[I]], align 4
// CHECK7-NEXT: store i32 456, ptr [[J]], align 4 // CHECK7-NEXT: store i32 456, ptr [[J]], align 4
// CHECK7-NEXT: [[A9:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 // CHECK7-NEXT: [[A9:%.*]] = getelementptr inbounds [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0
// CHECK7-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A9]], i32 0, i32 0 // CHECK7-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds [123 x [456 x i32]], ptr [[A9]], i32 0, i32 0
// CHECK7-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX10]], i32 0, i32 0 // CHECK7-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [456 x i32], ptr [[ARRAYIDX10]], i32 0, i32 0
// CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX11]], align 4 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[ARRAYIDX11]], align 4
Show All 13 Lines
// 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 ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [5 x i64], 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: [[_TMP1:%.*]] = 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_2:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i64, align 8
// CHECK9-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK9-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8 // CHECK9-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8
// CHECK9-NEXT: store i32 100, ptr [[N]], align 4 // CHECK9-NEXT: store i32 100, ptr [[N]], align 4
// CHECK9-NEXT: store i32 2, ptr [[M]], align 4 // CHECK9-NEXT: store i32 2, ptr [[M]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64
// CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[M]], align 4 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64 // CHECK9-NEXT: [[TMP3:%.*]] = zext i32 [[TMP2]] to i64
// CHECK9-NEXT: [[TMP4:%.*]] = call ptr @llvm.stacksave() // CHECK9-NEXT: [[TMP4:%.*]] = call ptr @llvm.stacksave()
// CHECK9-NEXT: store ptr [[TMP4]], ptr [[SAVED_STACK]], align 8 // CHECK9-NEXT: store ptr [[TMP4]], ptr [[SAVED_STACK]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]] // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], [[TMP3]]
// CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP5]], align 4 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP5]], align 4
// CHECK9-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8
// CHECK9-NEXT: store i64 [[TMP3]], ptr [[__VLA_EXPR1]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], ptr [[__VLA_EXPR1]], align 8
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[N]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP6]], ptr [[N_CASTED]], align 4 // CHECK9-NEXT: store i32 [[TMP6]], ptr [[N_CASTED]], align 4
// CHECK9-NEXT: [[TMP7:%.*]] = load i64, ptr [[N_CASTED]], align 8 // CHECK9-NEXT: [[TMP7:%.*]] = load i64, ptr [[N_CASTED]], align 8
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[M]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP8]], ptr [[M_CASTED]], align 4 // CHECK9-NEXT: store i32 [[TMP8]], ptr [[M_CASTED]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[M_CASTED]], align 8 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[M_CASTED]], align 8
// 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: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 40, i1 false) // CHECK9-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 40, i1 false)
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP12]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP13]], align 8 // CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP13]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store i64 [[TMP7]], ptr [[TMP15]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP14]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP15]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP16]], align 8
// CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK9-NEXT: store ptr null, ptr [[TMP17]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP17]], align 8
// CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP18]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP18]], align 8
// CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[TMP20]], align 8 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP19]], align 8
// CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK9-NEXT: store ptr null, ptr [[TMP22]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP20]], align 8
// CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP23]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP21]], align 8
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP25]], align 8 // CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP22]], align 8
// CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3
// CHECK9-NEXT: store ptr null, ptr [[TMP23]], align 8
// CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP24]], align 8
// CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP25]], align 8
// CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK9-NEXT: store i64 [[TMP11]], ptr [[TMP26]], align 8
// CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK9-NEXT: store ptr null, ptr [[TMP27]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP27]], align 8
// CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP28]], align 8 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: store i64 [[TMP3]], ptr [[TMP30]], align 8 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 // CHECK9-NEXT: store i32 [[TMP31]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: store ptr null, ptr [[TMP32]], align 8 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT: [[TMP33:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK9-NEXT: store i32 [[TMP32]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP33]], align 8 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP33]], 0
// CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP35]], align 8
// CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK9-NEXT: store i64 [[TMP11]], ptr [[TMP37]], align 8
// CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4
// CHECK9-NEXT: store ptr null, ptr [[TMP38]], align 8
// CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK9-NEXT: [[TMP42:%.*]] = load i32, ptr [[N]], align 4
// CHECK9-NEXT: store i32 [[TMP42]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP43:%.*]] = load i32, ptr [[M]], align 4
// CHECK9-NEXT: store i32 [[TMP43]], ptr [[DOTCAPTURE_EXPR_3]], align 4
// CHECK9-NEXT: [[TMP44:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP44]], 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: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP45:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_3]], align 4 // CHECK9-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP45]], 0 // CHECK9-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP34]], 0
// CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 // CHECK9-NEXT: [[DIV5:%.*]] = sdiv i32 [[SUB4]], 1
// CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64 // CHECK9-NEXT: [[CONV6:%.*]] = sext i32 [[DIV5]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV5]], [[CONV8]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV6]]
// CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB7:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB9]], ptr [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: store i64 [[SUB7]], ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK9-NEXT: [[TMP46:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_4]], align 8 // CHECK9-NEXT: [[TMP35:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP46]], 1 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP35]], 1
// CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK9-NEXT: store i32 1, ptr [[TMP47]], align 4 // CHECK9-NEXT: store i32 1, ptr [[TMP36]], align 4
// CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK9-NEXT: store i32 5, ptr [[TMP48]], align 4 // CHECK9-NEXT: store i32 5, ptr [[TMP37]], align 4
// CHECK9-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK9-NEXT: store ptr [[TMP39]], ptr [[TMP49]], align 8 // CHECK9-NEXT: store ptr [[TMP28]], ptr [[TMP38]], align 8
// CHECK9-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK9-NEXT: store ptr [[TMP40]], ptr [[TMP50]], align 8 // CHECK9-NEXT: store ptr [[TMP29]], ptr [[TMP39]], align 8
// CHECK9-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK9-NEXT: store ptr [[TMP41]], ptr [[TMP51]], align 8 // CHECK9-NEXT: store ptr [[TMP30]], ptr [[TMP40]], align 8
// CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK9-NEXT: store ptr @.offload_maptypes, ptr [[TMP52]], align 8 // CHECK9-NEXT: store ptr @.offload_maptypes, ptr [[TMP41]], align 8
// CHECK9-NEXT: [[TMP53:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK9-NEXT: store ptr null, ptr [[TMP53]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP42]], align 8
// CHECK9-NEXT: [[TMP54:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK9-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK9-NEXT: store ptr null, ptr [[TMP54]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP43]], align 8
// CHECK9-NEXT: [[TMP55:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK9-NEXT: store i64 [[ADD]], ptr [[TMP55]], align 8 // CHECK9-NEXT: store i64 [[ADD]], ptr [[TMP44]], align 8
// CHECK9-NEXT: [[TMP56:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, ptr [[KERNEL_ARGS]]) // CHECK9-NEXT: [[TMP45:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, ptr [[KERNEL_ARGS]])
// CHECK9-NEXT: [[TMP57:%.*]] = icmp ne i32 [[TMP56]], 0 // CHECK9-NEXT: [[TMP46:%.*]] = icmp ne i32 [[TMP45]], 0
// CHECK9-NEXT: br i1 [[TMP57]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP46]], 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]], ptr [[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]], ptr [[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: [[TMP58:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK9-NEXT: [[TMP47:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef signext [[TMP58]]) // CHECK9-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef signext [[TMP47]])
// CHECK9-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 // CHECK9-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4
// CHECK9-NEXT: [[TMP59:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 // CHECK9-NEXT: [[TMP48:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK9-NEXT: call void @llvm.stackrestore(ptr [[TMP59]]) // CHECK9-NEXT: call void @llvm.stackrestore(ptr [[TMP48]])
// CHECK9-NEXT: [[TMP60:%.*]] = load i32, ptr [[RETVAL]], align 4 // CHECK9-NEXT: [[TMP49:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK9-NEXT: ret i32 [[TMP60]] // CHECK9-NEXT: ret i32 [[TMP49]]
// //
// //
// 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 noundef [[N:%.*]], i64 noundef [[M:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[M:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef 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 All 26 Lines
// CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// 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
// CHECK9-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[VLA_ADDR2:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP3:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_5:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_4:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTCAPTURE_EXPR_6:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_5:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[J:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[J:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8
// CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[I13:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[I11:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[J14:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[J12:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8
// CHECK9-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8 // CHECK9-NEXT: store i64 [[M]], ptr [[M_ADDR]], align 8
// CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT: store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8 // CHECK9-NEXT: store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 // CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8
// CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_ADDR]], align 4
// CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[M_ADDR]], align 4
// CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0
// CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1
// CHECK9-NEXT: [[CONV7:%.*]] = sext i32 [[DIV]] to i64 // CHECK9-NEXT: [[CONV:%.*]] = sext i32 [[DIV]] to i64
// CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP6]], 0 // CHECK9-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP6]], 0
// CHECK9-NEXT: [[DIV9:%.*]] = sdiv i32 [[SUB8]], 1 // CHECK9-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1
// CHECK9-NEXT: [[CONV10:%.*]] = sext i32 [[DIV9]] to i64 // CHECK9-NEXT: [[CONV8:%.*]] = sext i32 [[DIV7]] to i64
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV7]], [[CONV10]] // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i64 [[CONV]], [[CONV8]]
// CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i64 [[MUL]], 1 // CHECK9-NEXT: [[SUB9:%.*]] = sub nsw i64 [[MUL]], 1
// CHECK9-NEXT: store i64 [[SUB11]], ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: store i64 [[SUB9]], ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: store i32 0, ptr [[I]], align 4 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4
// CHECK9-NEXT: store i32 0, ptr [[J]], align 4 // CHECK9-NEXT: store i32 0, ptr [[J]], align 4
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]] // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]]
// CHECK9-NEXT: br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]] // CHECK9-NEXT: br i1 [[CMP]], label [[LAND_LHS_TRUE:%.*]], label [[OMP_PRECOND_END:%.*]]
// CHECK9: land.lhs.true: // CHECK9: land.lhs.true:
// CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[CMP12:%.*]] = icmp slt i32 0, [[TMP8]] // CHECK9-NEXT: [[CMP10:%.*]] = icmp slt i32 0, [[TMP8]]
// CHECK9-NEXT: br i1 [[CMP12]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]] // CHECK9-NEXT: br i1 [[CMP10]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END]]
// CHECK9: omp.precond.then: // CHECK9: omp.precond.then:
// CHECK9-NEXT: store i64 0, ptr [[DOTOMP_LB]], align 8 // CHECK9-NEXT: store i64 0, ptr [[DOTOMP_LB]], align 8
// CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP9:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: store i64 [[TMP9]], ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: store i64 [[TMP9]], ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: store i64 1, ptr [[DOTOMP_STRIDE]], align 8 // CHECK9-NEXT: store i64 1, ptr [[DOTOMP_STRIDE]], align 8
// CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB1:[0-9]+]], i32 [[TMP11]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1) // CHECK9-NEXT: call void @__kmpc_for_static_init_8(ptr @[[GLOB1:[0-9]+]], i32 [[TMP11]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i64 1, i64 1)
// CHECK9-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: [[CMP15:%.*]] = icmp sgt i64 [[TMP12]], [[TMP13]] // CHECK9-NEXT: [[CMP13:%.*]] = icmp sgt i64 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: br i1 [[CMP15]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] // CHECK9-NEXT: br i1 [[CMP13]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK9: cond.true: // CHECK9: cond.true:
// CHECK9-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_6]], align 8 // CHECK9-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_5]], align 8
// CHECK9-NEXT: br label [[COND_END:%.*]] // CHECK9-NEXT: br label [[COND_END:%.*]]
// CHECK9: cond.false: // CHECK9: cond.false:
// CHECK9-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: br label [[COND_END]] // CHECK9-NEXT: br label [[COND_END]]
// CHECK9: cond.end: // CHECK9: cond.end:
// CHECK9-NEXT: [[COND:%.*]] = phi i64 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] // CHECK9-NEXT: [[COND:%.*]] = phi i64 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i64 [[COND]], ptr [[DOTOMP_UB]], align 8 // CHECK9-NEXT: store i64 [[COND]], ptr [[DOTOMP_UB]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8 // CHECK9-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK9-NEXT: store i64 [[TMP16]], ptr [[DOTOMP_IV]], align 8 // CHECK9-NEXT: store i64 [[TMP16]], ptr [[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: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !5 // CHECK9-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP5:![0-9]+]]
// CHECK9-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group !5 // CHECK9-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[CMP16:%.*]] = icmp sle i64 [[TMP17]], [[TMP18]] // CHECK9-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP17]], [[TMP18]]
// CHECK9-NEXT: br i1 [[CMP16]], 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: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !5 // CHECK9-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4, !llvm.access.group !5 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[SUB17:%.*]] = sub nsw i32 [[TMP20]], 0 // CHECK9-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP20]], 0
// CHECK9-NEXT: [[DIV18:%.*]] = sdiv i32 [[SUB17]], 1 // CHECK9-NEXT: [[DIV16:%.*]] = sdiv i32 [[SUB15]], 1
// CHECK9-NEXT: [[MUL19:%.*]] = mul nsw i32 1, [[DIV18]] // CHECK9-NEXT: [[MUL17:%.*]] = mul nsw i32 1, [[DIV16]]
// CHECK9-NEXT: [[CONV20:%.*]] = sext i32 [[MUL19]] to i64 // CHECK9-NEXT: [[CONV18:%.*]] = sext i32 [[MUL17]] to i64
// CHECK9-NEXT: [[DIV21:%.*]] = sdiv i64 [[TMP19]], [[CONV20]] // CHECK9-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP19]], [[CONV18]]
// CHECK9-NEXT: [[MUL22:%.*]] = mul nsw i64 [[DIV21]], 1 // CHECK9-NEXT: [[MUL20:%.*]] = mul nsw i64 [[DIV19]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL22]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL20]]
// CHECK9-NEXT: [[CONV23:%.*]] = trunc i64 [[ADD]] to i32 // CHECK9-NEXT: [[CONV21:%.*]] = trunc i64 [[ADD]] to i32
// CHECK9-NEXT: store i32 [[CONV23]], ptr [[I13]], align 4, !llvm.access.group !5 // CHECK9-NEXT: store i32 [[CONV21]], ptr [[I11]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !5 // CHECK9-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[TMP22:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !5 // CHECK9-NEXT: [[TMP22:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4, !llvm.access.group !5 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP23]], 0 // CHECK9-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP23]], 0
// CHECK9-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1 // CHECK9-NEXT: [[DIV23:%.*]] = sdiv i32 [[SUB22]], 1
// CHECK9-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]] // CHECK9-NEXT: [[MUL24:%.*]] = mul nsw i32 1, [[DIV23]]
// CHECK9-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64 // CHECK9-NEXT: [[CONV25:%.*]] = sext i32 [[MUL24]] to i64
// CHECK9-NEXT: [[DIV28:%.*]] = sdiv i64 [[TMP22]], [[CONV27]] // CHECK9-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP22]], [[CONV25]]
// CHECK9-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4, !llvm.access.group !5 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[SUB29:%.*]] = sub nsw i32 [[TMP24]], 0 // CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK9-NEXT: [[DIV30:%.*]] = sdiv i32 [[SUB29]], 1 // CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1
// CHECK9-NEXT: [[MUL31:%.*]] = mul nsw i32 1, [[DIV30]] // CHECK9-NEXT: [[MUL29:%.*]] = mul nsw i32 1, [[DIV28]]
// CHECK9-NEXT: [[CONV32:%.*]] = sext i32 [[MUL31]] to i64 // CHECK9-NEXT: [[CONV30:%.*]] = sext i32 [[MUL29]] to i64
// CHECK9-NEXT: [[MUL33:%.*]] = mul nsw i64 [[DIV28]], [[CONV32]] // CHECK9-NEXT: [[MUL31:%.*]] = mul nsw i64 [[DIV26]], [[CONV30]]
// CHECK9-NEXT: [[SUB34:%.*]] = sub nsw i64 [[TMP21]], [[MUL33]] // CHECK9-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP21]], [[MUL31]]
// CHECK9-NEXT: [[MUL35:%.*]] = mul nsw i64 [[SUB34]], 1 // CHECK9-NEXT: [[MUL33:%.*]] = mul nsw i64 [[SUB32]], 1
// CHECK9-NEXT: [[ADD36:%.*]] = add nsw i64 0, [[MUL35]] // CHECK9-NEXT: [[ADD34:%.*]] = add nsw i64 0, [[MUL33]]
// CHECK9-NEXT: [[CONV37:%.*]] = trunc i64 [[ADD36]] to i32 // CHECK9-NEXT: [[CONV35:%.*]] = trunc i64 [[ADD34]] to i32
// CHECK9-NEXT: store i32 [[CONV37]], ptr [[J14]], align 4, !llvm.access.group !5 // CHECK9-NEXT: store i32 [[CONV35]], ptr [[J12]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[TMP25:%.*]] = load i32, ptr [[I13]], align 4, !llvm.access.group !5 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, ptr [[I11]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP25]] to i64 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP25]] to i64
// CHECK9-NEXT: [[TMP26:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP1]] // CHECK9-NEXT: [[TMP26:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP1]]
// CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[TMP26]] // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[TMP26]]
// CHECK9-NEXT: [[TMP27:%.*]] = load i32, ptr [[J14]], align 4, !llvm.access.group !5 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, ptr [[J12]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[IDXPROM38:%.*]] = sext i32 [[TMP27]] to i64 // CHECK9-NEXT: [[IDXPROM36:%.*]] = sext i32 [[TMP27]] to i64
// CHECK9-NEXT: [[ARRAYIDX39:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i64 [[IDXPROM38]] // CHECK9-NEXT: [[ARRAYIDX37:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i64 [[IDXPROM36]]
// CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX39]], align 4, !llvm.access.group !5 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX37]], align 4, !llvm.access.group [[ACC_GRP5]]
// 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: [[TMP28:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !5 // CHECK9-NEXT: [[TMP28:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: [[ADD40:%.*]] = add nsw i64 [[TMP28]], 1 // CHECK9-NEXT: [[ADD38:%.*]] = add nsw i64 [[TMP28]], 1
// CHECK9-NEXT: store i64 [[ADD40]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group !5 // CHECK9-NEXT: store i64 [[ADD38]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP5]]
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![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: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK9-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK9-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4
// CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP30]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP30]])
// CHECK9-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 // CHECK9-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0
// CHECK9-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK9-NEXT: br i1 [[TMP32]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then: // CHECK9: .omp.final.then:
// CHECK9-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK9-NEXT: [[SUB41:%.*]] = sub nsw i32 [[TMP33]], 0 // CHECK9-NEXT: [[SUB39:%.*]] = sub nsw i32 [[TMP33]], 0
// CHECK9-NEXT: [[DIV42:%.*]] = sdiv i32 [[SUB41]], 1 // CHECK9-NEXT: [[DIV40:%.*]] = sdiv i32 [[SUB39]], 1
// CHECK9-NEXT: [[MUL43:%.*]] = mul nsw i32 [[DIV42]], 1 // CHECK9-NEXT: [[MUL41:%.*]] = mul nsw i32 [[DIV40]], 1
// CHECK9-NEXT: [[ADD44:%.*]] = add nsw i32 0, [[MUL43]] // CHECK9-NEXT: [[ADD42:%.*]] = add nsw i32 0, [[MUL41]]
// CHECK9-NEXT: store i32 [[ADD44]], ptr [[I13]], align 4 // CHECK9-NEXT: store i32 [[ADD42]], ptr [[I11]], align 4
// CHECK9-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_5]], align 4 // CHECK9-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4
// CHECK9-NEXT: [[SUB45:%.*]] = sub nsw i32 [[TMP34]], 0 // CHECK9-NEXT: [[SUB43:%.*]] = sub nsw i32 [[TMP34]], 0
// CHECK9-NEXT: [[DIV46:%.*]] = sdiv i32 [[SUB45]], 1 // CHECK9-NEXT: [[DIV44:%.*]] = sdiv i32 [[SUB43]], 1
// CHECK9-NEXT: [[MUL47:%.*]] = mul nsw i32 [[DIV46]], 1 // CHECK9-NEXT: [[MUL45:%.*]] = mul nsw i32 [[DIV44]], 1
// CHECK9-NEXT: [[ADD48:%.*]] = add nsw i32 0, [[MUL47]] // CHECK9-NEXT: [[ADD46:%.*]] = add nsw i32 0, [[MUL45]]
// CHECK9-NEXT: store i32 [[ADD48]], ptr [[J14]], align 4 // CHECK9-NEXT: store i32 [[ADD46]], ptr [[J12]], 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 {{[^@]+}}@_Z5tmainIiLi10ELi2EEiT_ // CHECK9-LABEL: define {{[^@]+}}@_Z5tmainIiLi10ELi2EEiT_
// CHECK9-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR6:[0-9]+]] comdat { // CHECK9-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR6:[0-9]+]] comdat {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[A:%.*]] = alloca [10 x [2 x i32]], align 4 // CHECK9-NEXT: [[A:%.*]] = alloca [10 x [2 x i32]], align 4
// CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[_TMP1:%.*]] = alloca i32, align 4
// CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK9-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: store ptr [[A]], ptr [[TMP0]], align 8 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP0]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: store ptr [[A]], ptr [[TMP2]], align 8 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP1]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK9-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK9-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK9-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK9-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK9-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK9-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8 // CHECK9-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK9-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK9-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK9-NEXT: store ptr @.offload_sizes.2, ptr [[TMP11]], align 8 // CHECK9-NEXT: store ptr @.offload_sizes.2, ptr [[TMP9]], align 8
// CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK9-NEXT: store ptr @.offload_maptypes.3, ptr [[TMP12]], align 8 // CHECK9-NEXT: store ptr @.offload_maptypes.3, ptr [[TMP10]], align 8
// CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK9-NEXT: store ptr null, ptr [[TMP13]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK9-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK9-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK9-NEXT: store i64 20, ptr [[TMP15]], align 8 // CHECK9-NEXT: store i64 20, ptr [[TMP13]], align 8
// CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, ptr [[KERNEL_ARGS]]) // CHECK9-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, ptr [[KERNEL_ARGS]])
// CHECK9-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK9-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK9-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK9-NEXT: br i1 [[TMP15]], 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(ptr [[A]]) #[[ATTR4]] // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67(ptr [[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
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67
▲ Show 20 Lines • Show All 42 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]], ptr [[DOTOMP_UB]], align 4 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[TMP5]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !11 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11:![0-9]+]]
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !11 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP11]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !11 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
// 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]], ptr [[I]], align 4, !llvm.access.group !11 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !11 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !11 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
// 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]], ptr [[J]], align 4, !llvm.access.group !11 // CHECK9-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !11 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP11]]
// 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]], ptr [[TMP0]], i64 0, i64 [[IDXPROM]] // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[TMP0]], i64 0, i64 [[IDXPROM]]
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group !11 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK9-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP12]] to i64 // CHECK9-NEXT: [[IDXPROM7:%.*]] = sext i32 [[TMP12]] to i64
// CHECK9-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]] // CHECK9-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM7]]
// CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX8]], align 4, !llvm.access.group !11 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX8]], align 4, !llvm.access.group [[ACC_GRP11]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !11 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK9-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP13]], 1 // CHECK9-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK9-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !11 // CHECK9-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[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:%.*]]
▲ Show 20 Lines • Show All 52 Lines • ▼ Show 20 Lines
// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[N_CASTED]], align 4 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[N_CASTED]], align 4
// CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[M]], align 4 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP6]], ptr [[M_CASTED]], align 4 // CHECK11-NEXT: store i32 [[TMP6]], ptr [[M_CASTED]], align 4
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[M_CASTED]], align 4 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[M_CASTED]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP0]], [[TMP1]] // CHECK11-NEXT: [[TMP8:%.*]] = mul nuw i32 [[TMP0]], [[TMP1]]
// 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: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 40, i1 false) // CHECK11-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 40, i1 false)
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP11]], align 4
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP12]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP12]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP5]], ptr [[TMP14]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP13]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP14]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP15]], align 4
// CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK11-NEXT: store ptr null, ptr [[TMP16]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP16]], align 4
// CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP17]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP17]], align 4
// CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK11-NEXT: store i32 [[TMP7]], ptr [[TMP19]], align 4 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP18]], align 4
// CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK11-NEXT: store ptr null, ptr [[TMP21]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP19]], align 4
// CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP22]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP20]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3
// CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP24]], align 4 // CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP21]], align 4
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3
// CHECK11-NEXT: store ptr null, ptr [[TMP22]], align 4
// CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4
// CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP23]], align 4
// CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4
// CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP24]], align 4
// CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK11-NEXT: store i64 [[TMP10]], ptr [[TMP25]], align 4
// CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK11-NEXT: store ptr null, ptr [[TMP26]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP26]], align 4
// CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP27]], align 4 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: store i32 [[TMP1]], ptr [[TMP29]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 // CHECK11-NEXT: store i32 [[TMP30]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: store ptr null, ptr [[TMP31]], align 4 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, ptr [[M]], align 4
// CHECK11-NEXT: [[TMP32:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 // CHECK11-NEXT: store i32 [[TMP31]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP32]], align 4 // CHECK11-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP32]], 0
// CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP34]], align 4
// CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 4
// CHECK11-NEXT: store i64 [[TMP10]], ptr [[TMP36]], align 4
// CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4
// CHECK11-NEXT: store ptr null, ptr [[TMP37]], align 4
// CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [5 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0
// CHECK11-NEXT: [[TMP41:%.*]] = load i32, ptr [[N]], align 4
// CHECK11-NEXT: store i32 [[TMP41]], ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[TMP42:%.*]] = load i32, ptr [[M]], align 4
// CHECK11-NEXT: store i32 [[TMP42]], ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[TMP43:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP43]], 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: [[TMP44:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK11-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP44]], 0 // CHECK11-NEXT: [[SUB4:%.*]] = sub nsw i32 [[TMP33]], 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]], ptr [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: store i64 [[SUB7]], ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[TMP45:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8 // CHECK11-NEXT: [[TMP34:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR_3]], align 8
// CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP45]], 1 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i64 [[TMP34]], 1
// CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK11-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK11-NEXT: store i32 1, ptr [[TMP46]], align 4 // CHECK11-NEXT: store i32 1, ptr [[TMP35]], align 4
// CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK11-NEXT: store i32 5, ptr [[TMP47]], align 4 // CHECK11-NEXT: store i32 5, ptr [[TMP36]], align 4
// CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK11-NEXT: store ptr [[TMP38]], ptr [[TMP48]], align 4 // CHECK11-NEXT: store ptr [[TMP27]], ptr [[TMP37]], align 4
// CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK11-NEXT: store ptr [[TMP39]], ptr [[TMP49]], align 4 // CHECK11-NEXT: store ptr [[TMP28]], ptr [[TMP38]], align 4
// CHECK11-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK11-NEXT: store ptr [[TMP40]], ptr [[TMP50]], align 4 // CHECK11-NEXT: store ptr [[TMP29]], ptr [[TMP39]], align 4
// CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK11-NEXT: store ptr @.offload_maptypes, ptr [[TMP51]], align 4 // CHECK11-NEXT: store ptr @.offload_maptypes, ptr [[TMP40]], align 4
// CHECK11-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK11-NEXT: store ptr null, ptr [[TMP52]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP41]], align 4
// CHECK11-NEXT: [[TMP53:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK11-NEXT: store ptr null, ptr [[TMP53]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP42]], align 4
// CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK11-NEXT: store i64 [[ADD]], ptr [[TMP54]], align 8 // CHECK11-NEXT: store i64 [[ADD]], ptr [[TMP43]], align 8
// CHECK11-NEXT: [[TMP55:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, ptr [[KERNEL_ARGS]]) // CHECK11-NEXT: [[TMP44:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l80.region_id, ptr [[KERNEL_ARGS]])
// CHECK11-NEXT: [[TMP56:%.*]] = icmp ne i32 [[TMP55]], 0 // CHECK11-NEXT: [[TMP45:%.*]] = icmp ne i32 [[TMP44]], 0
// CHECK11-NEXT: br i1 [[TMP56]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP45]], 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]], ptr [[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]], ptr [[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: [[TMP57:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 // CHECK11-NEXT: [[TMP46:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef [[TMP57]]) // CHECK11-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10ELi2EEiT_(i32 noundef [[TMP46]])
// CHECK11-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 // CHECK11-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4
// CHECK11-NEXT: [[TMP58:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4 // CHECK11-NEXT: [[TMP47:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4
// CHECK11-NEXT: call void @llvm.stackrestore(ptr [[TMP58]]) // CHECK11-NEXT: call void @llvm.stackrestore(ptr [[TMP47]])
// CHECK11-NEXT: [[TMP59:%.*]] = load i32, ptr [[RETVAL]], align 4 // CHECK11-NEXT: [[TMP48:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK11-NEXT: ret i32 [[TMP59]] // CHECK11-NEXT: ret i32 [[TMP48]]
// //
// //
// 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 noundef [[N:%.*]], i32 noundef [[M:%.*]], i32 noundef [[VLA:%.*]], i32 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[M:%.*]], i32 noundef [[VLA:%.*]], i32 noundef [[VLA1:%.*]], ptr noundef 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 98 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 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ] // CHECK11-NEXT: [[COND:%.*]] = phi i64 [ [[TMP14]], [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ]
// CHECK11-NEXT: store i64 [[COND]], ptr [[DOTOMP_UB]], align 8 // CHECK11-NEXT: store i64 [[COND]], ptr [[DOTOMP_UB]], align 8
// CHECK11-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8 // CHECK11-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK11-NEXT: store i64 [[TMP16]], ptr [[DOTOMP_IV]], align 8 // CHECK11-NEXT: store i64 [[TMP16]], ptr [[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: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !6 // CHECK11-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP6:![0-9]+]]
// CHECK11-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group !6 // CHECK11-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP17]], [[TMP18]] // CHECK11-NEXT: [[CMP14:%.*]] = icmp sle i64 [[TMP17]], [[TMP18]]
// 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: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !6 // CHECK11-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !6 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP20]], 0 // CHECK11-NEXT: [[SUB15:%.*]] = sub nsw i32 [[TMP20]], 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 [[TMP19]], [[CONV18]] // CHECK11-NEXT: [[DIV19:%.*]] = sdiv i64 [[TMP19]], [[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]], ptr [[I11]], align 4, !llvm.access.group !6 // CHECK11-NEXT: store i32 [[CONV21]], ptr [[I11]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !6 // CHECK11-NEXT: [[TMP21:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[TMP22:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !6 // CHECK11-NEXT: [[TMP22:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !6 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP23]], 0 // CHECK11-NEXT: [[SUB22:%.*]] = sub nsw i32 [[TMP23]], 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 [[TMP22]], [[CONV25]] // CHECK11-NEXT: [[DIV26:%.*]] = sdiv i64 [[TMP22]], [[CONV25]]
// CHECK11-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group !6 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_4]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP24]], 0 // CHECK11-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP24]], 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 [[TMP21]], [[MUL31]] // CHECK11-NEXT: [[SUB32:%.*]] = sub nsw i64 [[TMP21]], [[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]], ptr [[J12]], align 4, !llvm.access.group !6 // CHECK11-NEXT: store i32 [[CONV35]], ptr [[J12]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[TMP25:%.*]] = load i32, ptr [[I11]], align 4, !llvm.access.group !6 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, ptr [[I11]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[TMP26:%.*]] = mul nsw i32 [[TMP25]], [[TMP1]] // CHECK11-NEXT: [[TMP26:%.*]] = mul nsw i32 [[TMP25]], [[TMP1]]
// CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP26]] // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP26]]
// CHECK11-NEXT: [[TMP27:%.*]] = load i32, ptr [[J12]], align 4, !llvm.access.group !6 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, ptr [[J12]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[ARRAYIDX36:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i32 [[TMP27]] // CHECK11-NEXT: [[ARRAYIDX36:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i32 [[TMP27]]
// CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX36]], align 4, !llvm.access.group !6 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX36]], align 4, !llvm.access.group [[ACC_GRP6]]
// 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: [[TMP28:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !6 // CHECK11-NEXT: [[TMP28:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: [[ADD37:%.*]] = add nsw i64 [[TMP28]], 1 // CHECK11-NEXT: [[ADD37:%.*]] = add nsw i64 [[TMP28]], 1
// CHECK11-NEXT: store i64 [[ADD37]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group !6 // CHECK11-NEXT: store i64 [[ADD37]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP6]]
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![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: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK11-NEXT: [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK11-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4
// CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP30]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP30]])
// CHECK11-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
Show All 27 Lines
// CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], 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: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK11-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK11-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: store ptr [[A]], ptr [[TMP0]], align 4 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP0]], align 4
// CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: store ptr [[A]], ptr [[TMP2]], align 4 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP1]], align 4
// CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK11-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK11-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK11-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK11-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK11-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK11-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 // CHECK11-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK11-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK11-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK11-NEXT: store ptr @.offload_sizes.2, ptr [[TMP11]], align 4 // CHECK11-NEXT: store ptr @.offload_sizes.2, ptr [[TMP9]], align 4
// CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK11-NEXT: store ptr @.offload_maptypes.3, ptr [[TMP12]], align 4 // CHECK11-NEXT: store ptr @.offload_maptypes.3, ptr [[TMP10]], align 4
// CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK11-NEXT: store ptr null, ptr [[TMP13]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK11-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK11-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK11-NEXT: store i64 20, ptr [[TMP15]], align 8 // CHECK11-NEXT: store i64 20, ptr [[TMP13]], align 8
// CHECK11-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, ptr [[KERNEL_ARGS]]) // CHECK11-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67.region_id, ptr [[KERNEL_ARGS]])
// CHECK11-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK11-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK11-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK11-NEXT: br i1 [[TMP15]], 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(ptr [[A]]) #[[ATTR4]] // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67(ptr [[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
// //
// //
// CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10ELi2EEiT__l67
▲ Show 20 Lines • Show All 42 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]], ptr [[DOTOMP_UB]], align 4 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK11-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !12 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12:![0-9]+]]
// CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !12 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP12]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !12 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
// 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]], ptr [[I]], align 4, !llvm.access.group !12 // CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !12 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !12 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
// 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]], ptr [[J]], align 4, !llvm.access.group !12 // CHECK11-NEXT: store i32 [[ADD6]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !12 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[TMP0]], i32 0, i32 [[TMP11]] // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[TMP0]], i32 0, i32 [[TMP11]]
// CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group !12 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK11-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP12]] // CHECK11-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP12]]
// CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group !12 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group [[ACC_GRP12]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !12 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP13]], 1 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP13]], 1
// CHECK11-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !12 // CHECK11-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP13:![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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[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:%.*]]
▲ Show 20 Lines • Show All 77 Lines • ▼ Show 20 Lines
// CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK13-NEXT: [[CMP8:%.*]] = icmp slt i32 0, [[TMP12]] // CHECK13-NEXT: [[CMP8:%.*]] = icmp slt i32 0, [[TMP12]]
// CHECK13-NEXT: br i1 [[CMP8]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END]] // CHECK13-NEXT: br i1 [[CMP8]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END]]
// CHECK13: simd.if.then: // CHECK13: simd.if.then:
// CHECK13-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8 // CHECK13-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK13-NEXT: store i64 [[TMP13]], ptr [[DOTOMP_IV]], align 8 // CHECK13-NEXT: store i64 [[TMP13]], ptr [[DOTOMP_IV]], align 8
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK13: omp.inner.for.cond: // CHECK13: omp.inner.for.cond:
// CHECK13-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !2 // CHECK13-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2:![0-9]+]]
// CHECK13-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group !2 // CHECK13-NEXT: [[TMP15:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[CMP11:%.*]] = icmp sle i64 [[TMP14]], [[TMP15]] // CHECK13-NEXT: [[CMP11:%.*]] = icmp sle i64 [[TMP14]], [[TMP15]]
// CHECK13-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK13: omp.inner.for.body: // CHECK13: omp.inner.for.body:
// CHECK13-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !2 // CHECK13-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group !2 // CHECK13-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP17]], 0 // CHECK13-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP17]], 0
// CHECK13-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK13-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1
// CHECK13-NEXT: [[MUL14:%.*]] = mul nsw i32 1, [[DIV13]] // CHECK13-NEXT: [[MUL14:%.*]] = mul nsw i32 1, [[DIV13]]
// CHECK13-NEXT: [[CONV15:%.*]] = sext i32 [[MUL14]] to i64 // CHECK13-NEXT: [[CONV15:%.*]] = sext i32 [[MUL14]] to i64
// CHECK13-NEXT: [[DIV16:%.*]] = sdiv i64 [[TMP16]], [[CONV15]] // CHECK13-NEXT: [[DIV16:%.*]] = sdiv i64 [[TMP16]], [[CONV15]]
// CHECK13-NEXT: [[MUL17:%.*]] = mul nsw i64 [[DIV16]], 1 // CHECK13-NEXT: [[MUL17:%.*]] = mul nsw i64 [[DIV16]], 1
// CHECK13-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL17]] // CHECK13-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL17]]
// CHECK13-NEXT: [[CONV18:%.*]] = trunc i64 [[ADD]] to i32 // CHECK13-NEXT: [[CONV18:%.*]] = trunc i64 [[ADD]] to i32
// CHECK13-NEXT: store i32 [[CONV18]], ptr [[I9]], align 4, !llvm.access.group !2 // CHECK13-NEXT: store i32 [[CONV18]], ptr [[I9]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !2 // CHECK13-NEXT: [[TMP18:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !2 // CHECK13-NEXT: [[TMP19:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group !2 // CHECK13-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[SUB19:%.*]] = sub nsw i32 [[TMP20]], 0 // CHECK13-NEXT: [[SUB19:%.*]] = sub nsw i32 [[TMP20]], 0
// CHECK13-NEXT: [[DIV20:%.*]] = sdiv i32 [[SUB19]], 1 // CHECK13-NEXT: [[DIV20:%.*]] = sdiv i32 [[SUB19]], 1
// CHECK13-NEXT: [[MUL21:%.*]] = mul nsw i32 1, [[DIV20]] // CHECK13-NEXT: [[MUL21:%.*]] = mul nsw i32 1, [[DIV20]]
// CHECK13-NEXT: [[CONV22:%.*]] = sext i32 [[MUL21]] to i64 // CHECK13-NEXT: [[CONV22:%.*]] = sext i32 [[MUL21]] to i64
// CHECK13-NEXT: [[DIV23:%.*]] = sdiv i64 [[TMP19]], [[CONV22]] // CHECK13-NEXT: [[DIV23:%.*]] = sdiv i64 [[TMP19]], [[CONV22]]
// CHECK13-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group !2 // CHECK13-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP21]], 0 // CHECK13-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP21]], 0
// CHECK13-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1 // CHECK13-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1
// CHECK13-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]] // CHECK13-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]]
// CHECK13-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64 // CHECK13-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64
// CHECK13-NEXT: [[MUL28:%.*]] = mul nsw i64 [[DIV23]], [[CONV27]] // CHECK13-NEXT: [[MUL28:%.*]] = mul nsw i64 [[DIV23]], [[CONV27]]
// CHECK13-NEXT: [[SUB29:%.*]] = sub nsw i64 [[TMP18]], [[MUL28]] // CHECK13-NEXT: [[SUB29:%.*]] = sub nsw i64 [[TMP18]], [[MUL28]]
// CHECK13-NEXT: [[MUL30:%.*]] = mul nsw i64 [[SUB29]], 1 // CHECK13-NEXT: [[MUL30:%.*]] = mul nsw i64 [[SUB29]], 1
// CHECK13-NEXT: [[ADD31:%.*]] = add nsw i64 0, [[MUL30]] // CHECK13-NEXT: [[ADD31:%.*]] = add nsw i64 0, [[MUL30]]
// CHECK13-NEXT: [[CONV32:%.*]] = trunc i64 [[ADD31]] to i32 // CHECK13-NEXT: [[CONV32:%.*]] = trunc i64 [[ADD31]] to i32
// CHECK13-NEXT: store i32 [[CONV32]], ptr [[J10]], align 4, !llvm.access.group !2 // CHECK13-NEXT: store i32 [[CONV32]], ptr [[J10]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[TMP22:%.*]] = load i32, ptr [[I9]], align 4, !llvm.access.group !2 // CHECK13-NEXT: [[TMP22:%.*]] = load i32, ptr [[I9]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP22]] to i64 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP22]] to i64
// CHECK13-NEXT: [[TMP23:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP3]] // CHECK13-NEXT: [[TMP23:%.*]] = mul nsw i64 [[IDXPROM]], [[TMP3]]
// CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i64 [[TMP23]] // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i64 [[TMP23]]
// CHECK13-NEXT: [[TMP24:%.*]] = load i32, ptr [[J10]], align 4, !llvm.access.group !2 // CHECK13-NEXT: [[TMP24:%.*]] = load i32, ptr [[J10]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[IDXPROM33:%.*]] = sext i32 [[TMP24]] to i64 // CHECK13-NEXT: [[IDXPROM33:%.*]] = sext i32 [[TMP24]] to i64
// CHECK13-NEXT: [[ARRAYIDX34:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i64 [[IDXPROM33]] // CHECK13-NEXT: [[ARRAYIDX34:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i64 [[IDXPROM33]]
// CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX34]], align 4, !llvm.access.group !2 // CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX34]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK13: omp.body.continue: // CHECK13: omp.body.continue:
// CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK13: omp.inner.for.inc: // CHECK13: omp.inner.for.inc:
// CHECK13-NEXT: [[TMP25:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !2 // CHECK13-NEXT: [[TMP25:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: [[ADD35:%.*]] = add nsw i64 [[TMP25]], 1 // CHECK13-NEXT: [[ADD35:%.*]] = add nsw i64 [[TMP25]], 1
// CHECK13-NEXT: store i64 [[ADD35]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group !2 // CHECK13-NEXT: store i64 [[ADD35]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP2]]
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK13: omp.inner.for.end: // CHECK13: omp.inner.for.end:
// CHECK13-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK13-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK13-NEXT: [[SUB36:%.*]] = sub nsw i32 [[TMP26]], 0 // CHECK13-NEXT: [[SUB36:%.*]] = sub nsw i32 [[TMP26]], 0
// CHECK13-NEXT: [[DIV37:%.*]] = sdiv i32 [[SUB36]], 1 // CHECK13-NEXT: [[DIV37:%.*]] = sdiv i32 [[SUB36]], 1
// CHECK13-NEXT: [[MUL38:%.*]] = mul nsw i32 [[DIV37]], 1 // CHECK13-NEXT: [[MUL38:%.*]] = mul nsw i32 [[DIV37]], 1
// CHECK13-NEXT: [[ADD39:%.*]] = add nsw i32 0, [[MUL38]] // CHECK13-NEXT: [[ADD39:%.*]] = add nsw i32 0, [[MUL38]]
// CHECK13-NEXT: store i32 [[ADD39]], ptr [[I9]], align 4 // CHECK13-NEXT: store i32 [[ADD39]], ptr [[I9]], align 4
Show All 28 Lines
// CHECK13-NEXT: [[J:%.*]] = alloca i32, align 4 // CHECK13-NEXT: [[J:%.*]] = alloca i32, align 4
// CHECK13-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK13-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK13-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 19, ptr [[DOTOMP_UB]], align 4 // CHECK13-NEXT: store i32 19, ptr [[DOTOMP_UB]], align 4
// CHECK13-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK13-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK13-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 // CHECK13-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK13: omp.inner.for.cond: // CHECK13: omp.inner.for.cond:
// CHECK13-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK13-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6:![0-9]+]]
// CHECK13-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !6 // CHECK13-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] // CHECK13-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK13-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK13-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK13: omp.inner.for.body: // CHECK13: omp.inner.for.body:
// CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 2 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 2
// CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK13-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !6 // CHECK13-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[DIV2:%.*]] = sdiv i32 [[TMP5]], 2 // CHECK13-NEXT: [[DIV2:%.*]] = sdiv i32 [[TMP5]], 2
// CHECK13-NEXT: [[MUL3:%.*]] = mul nsw i32 [[DIV2]], 2 // CHECK13-NEXT: [[MUL3:%.*]] = mul nsw i32 [[DIV2]], 2
// CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL3]] // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL3]]
// CHECK13-NEXT: [[MUL4:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK13-NEXT: [[MUL4:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK13-NEXT: [[ADD5:%.*]] = add nsw i32 0, [[MUL4]] // CHECK13-NEXT: [[ADD5:%.*]] = add nsw i32 0, [[MUL4]]
// CHECK13-NEXT: store i32 [[ADD5]], ptr [[J]], align 4, !llvm.access.group !6 // CHECK13-NEXT: store i32 [[ADD5]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !6 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP6]] to i64 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP6]] to i64
// CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[A]], i64 0, i64 [[IDXPROM]] // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[A]], i64 0, i64 [[IDXPROM]]
// CHECK13-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group !6 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[IDXPROM6:%.*]] = sext i32 [[TMP7]] to i64 // CHECK13-NEXT: [[IDXPROM6:%.*]] = sext i32 [[TMP7]] to i64
// CHECK13-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM6]] // CHECK13-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i64 0, i64 [[IDXPROM6]]
// CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group !6 // CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX7]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK13: omp.body.continue: // CHECK13: omp.body.continue:
// CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK13: omp.inner.for.inc: // CHECK13: omp.inner.for.inc:
// CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP8]], 1 // CHECK13-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK13-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK13-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK13: omp.inner.for.end: // CHECK13: omp.inner.for.end:
// CHECK13-NEXT: store i32 10, ptr [[I]], align 4 // CHECK13-NEXT: store i32 10, ptr [[I]], align 4
// CHECK13-NEXT: store i32 2, ptr [[J]], align 4 // CHECK13-NEXT: store i32 2, ptr [[J]], align 4
// CHECK13-NEXT: ret i32 0 // CHECK13-NEXT: ret i32 0
// //
// //
// CHECK15-LABEL: define {{[^@]+}}@main // CHECK15-LABEL: define {{[^@]+}}@main
▲ Show 20 Lines • Show All 59 Lines • ▼ Show 20 Lines
// CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4
// CHECK15-NEXT: [[CMP8:%.*]] = icmp slt i32 0, [[TMP10]] // CHECK15-NEXT: [[CMP8:%.*]] = icmp slt i32 0, [[TMP10]]
// CHECK15-NEXT: br i1 [[CMP8]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END]] // CHECK15-NEXT: br i1 [[CMP8]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END]]
// CHECK15: simd.if.then: // CHECK15: simd.if.then:
// CHECK15-NEXT: [[TMP11:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8 // CHECK15-NEXT: [[TMP11:%.*]] = load i64, ptr [[DOTOMP_LB]], align 8
// CHECK15-NEXT: store i64 [[TMP11]], ptr [[DOTOMP_IV]], align 8 // CHECK15-NEXT: store i64 [[TMP11]], ptr [[DOTOMP_IV]], align 8
// CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK15: omp.inner.for.cond: // CHECK15: omp.inner.for.cond:
// CHECK15-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !3 // CHECK15-NEXT: [[TMP12:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3:![0-9]+]]
// CHECK15-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group !3 // CHECK15-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTOMP_UB]], align 8, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[CMP11:%.*]] = icmp sle i64 [[TMP12]], [[TMP13]] // CHECK15-NEXT: [[CMP11:%.*]] = icmp sle i64 [[TMP12]], [[TMP13]]
// CHECK15-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK15-NEXT: br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK15: omp.inner.for.body: // CHECK15: omp.inner.for.body:
// CHECK15-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !3 // CHECK15-NEXT: [[TMP14:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group !3 // CHECK15-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP15]], 0 // CHECK15-NEXT: [[SUB12:%.*]] = sub nsw i32 [[TMP15]], 0
// CHECK15-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1 // CHECK15-NEXT: [[DIV13:%.*]] = sdiv i32 [[SUB12]], 1
// CHECK15-NEXT: [[MUL14:%.*]] = mul nsw i32 1, [[DIV13]] // CHECK15-NEXT: [[MUL14:%.*]] = mul nsw i32 1, [[DIV13]]
// CHECK15-NEXT: [[CONV15:%.*]] = sext i32 [[MUL14]] to i64 // CHECK15-NEXT: [[CONV15:%.*]] = sext i32 [[MUL14]] to i64
// CHECK15-NEXT: [[DIV16:%.*]] = sdiv i64 [[TMP14]], [[CONV15]] // CHECK15-NEXT: [[DIV16:%.*]] = sdiv i64 [[TMP14]], [[CONV15]]
// CHECK15-NEXT: [[MUL17:%.*]] = mul nsw i64 [[DIV16]], 1 // CHECK15-NEXT: [[MUL17:%.*]] = mul nsw i64 [[DIV16]], 1
// CHECK15-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL17]] // CHECK15-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL17]]
// CHECK15-NEXT: [[CONV18:%.*]] = trunc i64 [[ADD]] to i32 // CHECK15-NEXT: [[CONV18:%.*]] = trunc i64 [[ADD]] to i32
// CHECK15-NEXT: store i32 [[CONV18]], ptr [[I9]], align 4, !llvm.access.group !3 // CHECK15-NEXT: store i32 [[CONV18]], ptr [[I9]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !3 // CHECK15-NEXT: [[TMP16:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !3 // CHECK15-NEXT: [[TMP17:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group !3 // CHECK15-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[SUB19:%.*]] = sub nsw i32 [[TMP18]], 0 // CHECK15-NEXT: [[SUB19:%.*]] = sub nsw i32 [[TMP18]], 0
// CHECK15-NEXT: [[DIV20:%.*]] = sdiv i32 [[SUB19]], 1 // CHECK15-NEXT: [[DIV20:%.*]] = sdiv i32 [[SUB19]], 1
// CHECK15-NEXT: [[MUL21:%.*]] = mul nsw i32 1, [[DIV20]] // CHECK15-NEXT: [[MUL21:%.*]] = mul nsw i32 1, [[DIV20]]
// CHECK15-NEXT: [[CONV22:%.*]] = sext i32 [[MUL21]] to i64 // CHECK15-NEXT: [[CONV22:%.*]] = sext i32 [[MUL21]] to i64
// CHECK15-NEXT: [[DIV23:%.*]] = sdiv i64 [[TMP17]], [[CONV22]] // CHECK15-NEXT: [[DIV23:%.*]] = sdiv i64 [[TMP17]], [[CONV22]]
// CHECK15-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group !3 // CHECK15-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP19]], 0 // CHECK15-NEXT: [[SUB24:%.*]] = sub nsw i32 [[TMP19]], 0
// CHECK15-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1 // CHECK15-NEXT: [[DIV25:%.*]] = sdiv i32 [[SUB24]], 1
// CHECK15-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]] // CHECK15-NEXT: [[MUL26:%.*]] = mul nsw i32 1, [[DIV25]]
// CHECK15-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64 // CHECK15-NEXT: [[CONV27:%.*]] = sext i32 [[MUL26]] to i64
// CHECK15-NEXT: [[MUL28:%.*]] = mul nsw i64 [[DIV23]], [[CONV27]] // CHECK15-NEXT: [[MUL28:%.*]] = mul nsw i64 [[DIV23]], [[CONV27]]
// CHECK15-NEXT: [[SUB29:%.*]] = sub nsw i64 [[TMP16]], [[MUL28]] // CHECK15-NEXT: [[SUB29:%.*]] = sub nsw i64 [[TMP16]], [[MUL28]]
// CHECK15-NEXT: [[MUL30:%.*]] = mul nsw i64 [[SUB29]], 1 // CHECK15-NEXT: [[MUL30:%.*]] = mul nsw i64 [[SUB29]], 1
// CHECK15-NEXT: [[ADD31:%.*]] = add nsw i64 0, [[MUL30]] // CHECK15-NEXT: [[ADD31:%.*]] = add nsw i64 0, [[MUL30]]
// CHECK15-NEXT: [[CONV32:%.*]] = trunc i64 [[ADD31]] to i32 // CHECK15-NEXT: [[CONV32:%.*]] = trunc i64 [[ADD31]] to i32
// CHECK15-NEXT: store i32 [[CONV32]], ptr [[J10]], align 4, !llvm.access.group !3 // CHECK15-NEXT: store i32 [[CONV32]], ptr [[J10]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[TMP20:%.*]] = load i32, ptr [[I9]], align 4, !llvm.access.group !3 // CHECK15-NEXT: [[TMP20:%.*]] = load i32, ptr [[I9]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[TMP21:%.*]] = mul nsw i32 [[TMP20]], [[TMP1]] // CHECK15-NEXT: [[TMP21:%.*]] = mul nsw i32 [[TMP20]], [[TMP1]]
// CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i32 [[TMP21]] // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i32 [[TMP21]]
// CHECK15-NEXT: [[TMP22:%.*]] = load i32, ptr [[J10]], align 4, !llvm.access.group !3 // CHECK15-NEXT: [[TMP22:%.*]] = load i32, ptr [[J10]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[ARRAYIDX33:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i32 [[TMP22]] // CHECK15-NEXT: [[ARRAYIDX33:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i32 [[TMP22]]
// CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX33]], align 4, !llvm.access.group !3 // CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX33]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK15: omp.body.continue: // CHECK15: omp.body.continue:
// CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK15: omp.inner.for.inc: // CHECK15: omp.inner.for.inc:
// CHECK15-NEXT: [[TMP23:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group !3 // CHECK15-NEXT: [[TMP23:%.*]] = load i64, ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: [[ADD34:%.*]] = add nsw i64 [[TMP23]], 1 // CHECK15-NEXT: [[ADD34:%.*]] = add nsw i64 [[TMP23]], 1
// CHECK15-NEXT: store i64 [[ADD34]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group !3 // CHECK15-NEXT: store i64 [[ADD34]], ptr [[DOTOMP_IV]], align 8, !llvm.access.group [[ACC_GRP3]]
// CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
// CHECK15: omp.inner.for.end: // CHECK15: omp.inner.for.end:
// CHECK15-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 // CHECK15-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4
// CHECK15-NEXT: [[SUB35:%.*]] = sub nsw i32 [[TMP24]], 0 // CHECK15-NEXT: [[SUB35:%.*]] = sub nsw i32 [[TMP24]], 0
// CHECK15-NEXT: [[DIV36:%.*]] = sdiv i32 [[SUB35]], 1 // CHECK15-NEXT: [[DIV36:%.*]] = sdiv i32 [[SUB35]], 1
// CHECK15-NEXT: [[MUL37:%.*]] = mul nsw i32 [[DIV36]], 1 // CHECK15-NEXT: [[MUL37:%.*]] = mul nsw i32 [[DIV36]], 1
// CHECK15-NEXT: [[ADD38:%.*]] = add nsw i32 0, [[MUL37]] // CHECK15-NEXT: [[ADD38:%.*]] = add nsw i32 0, [[MUL37]]
// CHECK15-NEXT: store i32 [[ADD38]], ptr [[I9]], align 4 // CHECK15-NEXT: store i32 [[ADD38]], ptr [[I9]], align 4
Show All 28 Lines
// CHECK15-NEXT: [[J:%.*]] = alloca i32, align 4 // CHECK15-NEXT: [[J:%.*]] = alloca i32, align 4
// CHECK15-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 // CHECK15-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4
// CHECK15-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK15-NEXT: store i32 19, ptr [[DOTOMP_UB]], align 4 // CHECK15-NEXT: store i32 19, ptr [[DOTOMP_UB]], align 4
// CHECK15-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK15-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK15-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 // CHECK15-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK15: omp.inner.for.cond: // CHECK15: omp.inner.for.cond:
// CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK15-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7:![0-9]+]]
// CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !7 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] // CHECK15-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK15-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK15-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK15: omp.inner.for.body: // CHECK15: omp.inner.for.body:
// CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 2 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP3]], 2
// CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[DIV]], 1
// CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK15-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !7 // CHECK15-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[DIV2:%.*]] = sdiv i32 [[TMP5]], 2 // CHECK15-NEXT: [[DIV2:%.*]] = sdiv i32 [[TMP5]], 2
// CHECK15-NEXT: [[MUL3:%.*]] = mul nsw i32 [[DIV2]], 2 // CHECK15-NEXT: [[MUL3:%.*]] = mul nsw i32 [[DIV2]], 2
// CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL3]] // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], [[MUL3]]
// CHECK15-NEXT: [[MUL4:%.*]] = mul nsw i32 [[SUB]], 1 // CHECK15-NEXT: [[MUL4:%.*]] = mul nsw i32 [[SUB]], 1
// CHECK15-NEXT: [[ADD5:%.*]] = add nsw i32 0, [[MUL4]] // CHECK15-NEXT: [[ADD5:%.*]] = add nsw i32 0, [[MUL4]]
// CHECK15-NEXT: store i32 [[ADD5]], ptr [[J]], align 4, !llvm.access.group !7 // CHECK15-NEXT: store i32 [[ADD5]], ptr [[J]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !7 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[A]], i32 0, i32 [[TMP6]] // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x [2 x i32]], ptr [[A]], i32 0, i32 [[TMP6]]
// CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group !7 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[J]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP7]] // CHECK15-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x i32], ptr [[ARRAYIDX]], i32 0, i32 [[TMP7]]
// CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX6]], align 4, !llvm.access.group !7 // CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX6]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK15: omp.body.continue: // CHECK15: omp.body.continue:
// CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK15: omp.inner.for.inc: // CHECK15: omp.inner.for.inc:
// CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP8]], 1 // CHECK15-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP8]], 1
// CHECK15-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK15-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]] // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK15: omp.inner.for.end: // CHECK15: omp.inner.for.end:
// CHECK15-NEXT: store i32 10, ptr [[I]], align 4 // CHECK15-NEXT: store i32 10, ptr [[I]], align 4
// CHECK15-NEXT: store i32 2, ptr [[J]], align 4 // CHECK15-NEXT: store i32 2, ptr [[J]], align 4
// CHECK15-NEXT: ret i32 0 // CHECK15-NEXT: ret i32 0
// //

clang/test/OpenMP/target_teams_distribute_simd_reduction_codegen.cpp

Show First 20 Lines • Show All 83 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK1-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 8 // CHECK1-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 2, ptr [[TMP15]], align 8 // CHECK1-NEXT: store i64 2, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l60.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l60.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP15]], 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]+}}_main_l60(ptr @_ZZ4mainE5sivar) #[[ATTR3:[0-9]+]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l60(ptr @_ZZ4mainE5sivar) #[[ATTR3:[0-9]+]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v() // CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v()
// CHECK1-NEXT: ret i32 [[CALL]] // CHECK1-NEXT: ret i32 [[CALL]]
// //
// //
▲ Show 20 Lines • Show All 44 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 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !5 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5:![0-9]+]]
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !5 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP5]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !5 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !5 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !5 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[SIVAR1]], align 4, !llvm.access.group !5 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[SIVAR1]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]] // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]]
// CHECK1-NEXT: store i32 [[ADD3]], ptr [[SIVAR1]], align 4, !llvm.access.group !5 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[SIVAR1]], align 4, !llvm.access.group [[ACC_GRP5]]
// 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: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !5 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1 // CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1
// CHECK1-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !5 // CHECK1-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK1-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK1-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK1-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK1: .omp.final.then: // CHECK1: .omp.final.then:
// CHECK1-NEXT: store i32 2, ptr [[I]], align 4 // CHECK1-NEXT: store i32 2, ptr [[I]], align 4
// CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK1: .omp.final.done: // CHECK1: .omp.final.done:
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[SIVAR1]], ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr [[SIVAR1]], ptr [[TMP14]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK1-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK1-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4 // CHECK1-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4:[0-9]+]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v // CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK1-SAME: () #[[ATTR6:[0-9]+]] comdat { // CHECK1-SAME: () #[[ATTR6:[0-9]+]] comdat {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[T_VAR:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4 // CHECK1-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT: store i32 0, ptr [[T_VAR]], align 4 // CHECK1-NEXT: store i32 0, ptr [[T_VAR]], align 4
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i64 8, i1 false) // CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i64 8, i1 false)
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[T_VAR]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[T_VAR]], ptr [[TMP1]], align 8 // CHECK1-NEXT: store ptr [[T_VAR]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[T_VAR]], ptr [[TMP3]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr null, ptr [[TMP5]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP9]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 8 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.3, ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.3, ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP13]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP15]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 2, ptr [[TMP16]], align 8 // CHECK1-NEXT: store i64 2, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP18]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP15]], 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]+}}__Z5tmainIiET_v_l32(ptr [[T_VAR]]) #[[ATTR3]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32(ptr [[T_VAR]]) #[[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: ret i32 0 // CHECK1-NEXT: ret i32 0
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32
▲ Show 20 Lines • Show All 43 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 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !11 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11:![0-9]+]]
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !11 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP11]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !11 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !11 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !11 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group !11 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]] // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]]
// CHECK1-NEXT: store i32 [[ADD3]], ptr [[T_VAR1]], align 4, !llvm.access.group !11 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP11]]
// 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: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !11 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1 // CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1
// CHECK1-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !11 // CHECK1-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]]
// CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]] // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 // CHECK1-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0
// CHECK1-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK1-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK1: .omp.final.then: // CHECK1: .omp.final.then:
// CHECK1-NEXT: store i32 2, ptr [[I]], align 4 // CHECK1-NEXT: store i32 2, ptr [[I]], align 4
// CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK1: .omp.final.done: // CHECK1: .omp.final.done:
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[T_VAR1]], ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr [[T_VAR1]], ptr [[TMP14]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK1-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK1-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4 // CHECK1-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK1-SAME: () #[[ATTR8:[0-9]+]] { // CHECK1-SAME: () #[[ATTR8:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: call void @__tgt_register_requires(i64 1) // CHECK1-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@main // CHECK3-LABEL: define {{[^@]+}}@main
// CHECK3-SAME: () #[[ATTR0:[0-9]+]] { // CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK3-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 4 // CHECK3-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP2]], align 4 // CHECK3-NEXT: store ptr @_ZZ4mainE5sivar, ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP11]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 2, ptr [[TMP15]], align 8 // CHECK3-NEXT: store i64 2, ptr [[TMP13]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l60.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l60.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP15]], 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]+}}_main_l60(ptr @_ZZ4mainE5sivar) #[[ATTR3:[0-9]+]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l60(ptr @_ZZ4mainE5sivar) #[[ATTR3:[0-9]+]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v() // CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v()
// CHECK3-NEXT: ret i32 [[CALL]] // CHECK3-NEXT: ret i32 [[CALL]]
// //
// //
▲ Show 20 Lines • Show All 44 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 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6:![0-9]+]]
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !6 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP6]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !6 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !6 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[SIVAR1]], align 4, !llvm.access.group !6 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[SIVAR1]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]] // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]]
// CHECK3-NEXT: store i32 [[ADD3]], ptr [[SIVAR1]], align 4, !llvm.access.group !6 // CHECK3-NEXT: store i32 [[ADD3]], ptr [[SIVAR1]], align 4, !llvm.access.group [[ACC_GRP6]]
// 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: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1 // CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1
// CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[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:
// CHECK3-NEXT: store i32 2, ptr [[I]], align 4 // CHECK3-NEXT: store i32 2, ptr [[I]], align 4
// CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK3: .omp.final.done: // CHECK3: .omp.final.done:
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[SIVAR1]], ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr [[SIVAR1]], ptr [[TMP14]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK3-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK3-NEXT: ] // CHECK3-NEXT: ]
// CHECK3: .omp.reduction.case1: // CHECK3: .omp.reduction.case1:
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.case2: // CHECK3: .omp.reduction.case2:
// CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK3-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4 // CHECK3-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.default: // CHECK3: .omp.reduction.default:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4:[0-9]+]] { // CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@_Z5tmainIiET_v // CHECK3-LABEL: define {{[^@]+}}@_Z5tmainIiET_v
// CHECK3-SAME: () #[[ATTR6:[0-9]+]] comdat { // CHECK3-SAME: () #[[ATTR6:[0-9]+]] comdat {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[T_VAR:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[T_VAR:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4 // CHECK3-NEXT: [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK3-NEXT: store i32 0, ptr [[T_VAR]], align 4 // CHECK3-NEXT: store i32 0, ptr [[T_VAR]], align 4
// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i32 8, i1 false) // CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i32 8, i1 false)
// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR]], ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR]], ptr [[TMP1]], align 4 // CHECK3-NEXT: store ptr [[T_VAR]], ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR]], ptr [[TMP3]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP5]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, ptr [[TMP9]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 4 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes.3, ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes.3, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP13]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP15]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 2, ptr [[TMP16]], align 8 // CHECK3-NEXT: store i64 2, ptr [[TMP13]], align 8
// CHECK3-NEXT: [[TMP17:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP18]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP15]], 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]+}}__Z5tmainIiET_v_l32(ptr [[T_VAR]]) #[[ATTR3]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32(ptr [[T_VAR]]) #[[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: ret i32 0 // CHECK3-NEXT: ret i32 0
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiET_v_l32
▲ Show 20 Lines • Show All 43 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 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !12 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12:![0-9]+]]
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !12 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP12]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !12 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !12 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !12 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group !12 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]] // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]]
// CHECK3-NEXT: store i32 [[ADD3]], ptr [[T_VAR1]], align 4, !llvm.access.group !12 // CHECK3-NEXT: store i32 [[ADD3]], ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP12]]
// 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: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !12 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1 // CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP11]], 1
// CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !12 // CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]]
// CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP13:![0-9]+]] // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP13:![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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[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:
// CHECK3-NEXT: store i32 2, ptr [[I]], align 4 // CHECK3-NEXT: store i32 2, ptr [[I]], align 4
// CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]]
// CHECK3: .omp.final.done: // CHECK3: .omp.final.done:
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[T_VAR1]], ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr [[T_VAR1]], ptr [[TMP14]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP2]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: [[TMP15:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP2]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.2, ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: switch i32 [[TMP17]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK3-NEXT: switch i32 [[TMP15]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK3-NEXT: ] // CHECK3-NEXT: ]
// CHECK3: .omp.reduction.case1: // CHECK3: .omp.reduction.case1:
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.case2: // CHECK3: .omp.reduction.case2:
// CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK3-NEXT: [[TMP21:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP20]] monotonic, align 4 // CHECK3-NEXT: [[TMP19:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP18]] monotonic, align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.default: // CHECK3: .omp.reduction.default:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2 // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.2
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4]] { // CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR4]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK3-SAME: () #[[ATTR8:[0-9]+]] { // CHECK3-SAME: () #[[ATTR8:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: call void @__tgt_register_requires(i64 1) // CHECK3-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
Show All 12 Lines
// CHECK5-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK5-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4 // CHECK5-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: store i32 0, ptr [[SIVAR]], align 4 // CHECK5-NEXT: store i32 0, ptr [[SIVAR]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond: // CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2:![0-9]+]]
// CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] // CHECK5-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK5-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body: // CHECK5: omp.inner.for.body:
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !2 // CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[SIVAR]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[SIVAR]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], [[TMP4]] // CHECK5-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], [[TMP4]]
// CHECK5-NEXT: store i32 [[ADD1]], ptr [[SIVAR]], align 4, !llvm.access.group !2 // CHECK5-NEXT: store i32 [[ADD1]], ptr [[SIVAR]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK5: omp.body.continue: // CHECK5: omp.body.continue:
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc: // CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], 1 // CHECK5-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK5-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !2 // CHECK5-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]]
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]]
// CHECK5: omp.inner.for.end: // CHECK5: omp.inner.for.end:
// CHECK5-NEXT: store i32 2, ptr [[I]], align 4 // CHECK5-NEXT: store i32 2, ptr [[I]], align 4
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[SIVAR]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[SIVAR]], align 4
// CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP7]], [[TMP8]] // CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
// CHECK5-NEXT: store i32 [[ADD3]], ptr @_ZZ4mainE5sivar, align 4 // CHECK5-NEXT: store i32 [[ADD3]], ptr @_ZZ4mainE5sivar, align 4
// CHECK5-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v() // CHECK5-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiET_v()
Show All 10 Lines
// CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK5-NEXT: [[T_VAR1:%.*]] = alloca i32, align 4 // CHECK5-NEXT: [[T_VAR1:%.*]] = alloca i32, align 4
// CHECK5-NEXT: store i32 0, ptr [[T_VAR]], align 4 // CHECK5-NEXT: store i32 0, ptr [[T_VAR]], align 4
// CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i64 8, i1 false) // CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i64 8, i1 false)
// CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4 // CHECK5-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK5-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK5-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK5-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_IV]], align 4 // CHECK5-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK5-NEXT: store i32 0, ptr [[T_VAR1]], align 4 // CHECK5-NEXT: store i32 0, ptr [[T_VAR1]], align 4
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK5: omp.inner.for.cond: // CHECK5: omp.inner.for.cond:
// CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6:![0-9]+]]
// CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !6 // CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK5-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP2]], [[TMP3]] // CHECK5-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK5-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK5-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK5: omp.inner.for.body: // CHECK5: omp.inner.for.body:
// CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP4]], 1 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !6 // CHECK5-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !6 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group !6 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK5-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], [[TMP5]] // CHECK5-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP5]], [[TMP4]]
// CHECK5-NEXT: store i32 [[ADD2]], ptr [[T_VAR1]], align 4, !llvm.access.group !6 // CHECK5-NEXT: store i32 [[ADD2]], ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK5: omp.body.continue: // CHECK5: omp.body.continue:
// CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK5: omp.inner.for.inc: // CHECK5: omp.inner.for.inc:
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP7]], 1 // CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK5-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !6 // CHECK5-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]]
// CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]]
// CHECK5: omp.inner.for.end: // CHECK5: omp.inner.for.end:
// CHECK5-NEXT: store i32 2, ptr [[I]], align 4 // CHECK5-NEXT: store i32 2, ptr [[I]], align 4
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[T_VAR]], align 4 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[T_VAR]], align 4
// CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK5-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP8]], [[TMP9]] // CHECK5-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
// CHECK5-NEXT: store i32 [[ADD4]], ptr [[T_VAR]], align 4 // CHECK5-NEXT: store i32 [[ADD4]], ptr [[T_VAR]], align 4
// CHECK5-NEXT: ret i32 0 // CHECK5-NEXT: ret i32 0
// //
// //
// CHECK7-LABEL: define {{[^@]+}}@main // CHECK7-LABEL: define {{[^@]+}}@main
// CHECK7-SAME: () #[[ATTR0:[0-9]+]] { // CHECK7-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK7-NEXT: entry: // CHECK7-NEXT: entry:
// CHECK7-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[SIVAR:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[SIVAR:%.*]] = alloca i32, align 4
// CHECK7-NEXT: store i32 0, ptr [[RETVAL]], align 4 // CHECK7-NEXT: store i32 0, ptr [[RETVAL]], align 4
// CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK7-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4 // CHECK7-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK7-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK7-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK7-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 // CHECK7-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK7-NEXT: store i32 0, ptr [[SIVAR]], align 4 // CHECK7-NEXT: store i32 0, ptr [[SIVAR]], align 4
// CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK7: omp.inner.for.cond: // CHECK7: omp.inner.for.cond:
// CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3:![0-9]+]]
// CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] // CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK7: omp.inner.for.body: // CHECK7: omp.inner.for.body:
// CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 // CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !3 // CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[SIVAR]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[SIVAR]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], [[TMP4]] // CHECK7-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], [[TMP4]]
// CHECK7-NEXT: store i32 [[ADD1]], ptr [[SIVAR]], align 4, !llvm.access.group !3 // CHECK7-NEXT: store i32 [[ADD1]], ptr [[SIVAR]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK7: omp.body.continue: // CHECK7: omp.body.continue:
// CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK7: omp.inner.for.inc: // CHECK7: omp.inner.for.inc:
// CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], 1 // CHECK7-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK7-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !3 // CHECK7-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]]
// CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]]
// CHECK7: omp.inner.for.end: // CHECK7: omp.inner.for.end:
// CHECK7-NEXT: store i32 2, ptr [[I]], align 4 // CHECK7-NEXT: store i32 2, ptr [[I]], align 4
// CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr @_ZZ4mainE5sivar, align 4
// CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[SIVAR]], align 4 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[SIVAR]], align 4
// CHECK7-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP7]], [[TMP8]] // CHECK7-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
// CHECK7-NEXT: store i32 [[ADD3]], ptr @_ZZ4mainE5sivar, align 4 // CHECK7-NEXT: store i32 [[ADD3]], ptr @_ZZ4mainE5sivar, align 4
// CHECK7-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v() // CHECK7-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiET_v()
Show All 10 Lines
// CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4
// CHECK7-NEXT: [[T_VAR1:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[T_VAR1:%.*]] = alloca i32, align 4
// CHECK7-NEXT: store i32 0, ptr [[T_VAR]], align 4 // CHECK7-NEXT: store i32 0, ptr [[T_VAR]], align 4
// CHECK7-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i32 8, i1 false) // CHECK7-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiET_v.vec, i32 8, i1 false)
// CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK7-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4 // CHECK7-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK7-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK7-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_IV]], align 4 // CHECK7-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4
// CHECK7-NEXT: store i32 0, ptr [[T_VAR1]], align 4 // CHECK7-NEXT: store i32 0, ptr [[T_VAR1]], align 4
// CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK7: omp.inner.for.cond: // CHECK7: omp.inner.for.cond:
// CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7:![0-9]+]]
// CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !7 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP2]], [[TMP3]] // CHECK7-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]]
// CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] // CHECK7-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK7: omp.inner.for.body: // CHECK7: omp.inner.for.body:
// CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP4]], 1 // CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !7 // CHECK7-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !7 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group !7 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK7-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], [[TMP5]] // CHECK7-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP5]], [[TMP4]]
// CHECK7-NEXT: store i32 [[ADD2]], ptr [[T_VAR1]], align 4, !llvm.access.group !7 // CHECK7-NEXT: store i32 [[ADD2]], ptr [[T_VAR1]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK7: omp.body.continue: // CHECK7: omp.body.continue:
// CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK7: omp.inner.for.inc: // CHECK7: omp.inner.for.inc:
// CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK7-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP7]], 1 // CHECK7-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP6]], 1
// CHECK7-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !7 // CHECK7-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]]
// CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]] // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]]
// CHECK7: omp.inner.for.end: // CHECK7: omp.inner.for.end:
// CHECK7-NEXT: store i32 2, ptr [[I]], align 4 // CHECK7-NEXT: store i32 2, ptr [[I]], align 4
// CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[T_VAR]], align 4 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[T_VAR]], align 4
// CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[T_VAR1]], align 4 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[T_VAR1]], align 4
// CHECK7-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP8]], [[TMP9]] // CHECK7-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
// CHECK7-NEXT: store i32 [[ADD4]], ptr [[T_VAR]], align 4 // CHECK7-NEXT: store i32 [[ADD4]], ptr [[T_VAR]], align 4
// CHECK7-NEXT: ret i32 0 // CHECK7-NEXT: ret i32 0
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@main // CHECK9-LABEL: define {{[^@]+}}@main
// CHECK9-SAME: () #[[ATTR0:[0-9]+]] { // CHECK9-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 // CHECK9-NEXT: [[RETVAL:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 51 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 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ]
// CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK9-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 // CHECK9-NEXT: store i32 [[TMP5]], ptr [[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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4:![0-9]+]]
// CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group !4 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP4]]
// 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, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group !4 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group !4 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[SIVAR1]], align 4, !llvm.access.group !4 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[SIVAR1]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]] // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP10]], [[TMP9]]
// CHECK9-NEXT: store i32 [[ADD3]], ptr [[SIVAR1]], align 4, !llvm.access.group !4 // CHECK9-NEXT: store i32 [[ADD3]], ptr [[SIVAR1]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[REF_TMP]], i32 0, i32 0 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[CLASS_ANON_0]], ptr [[REF_TMP]], i32 0, i32 0
// CHECK9-NEXT: store ptr [[SIVAR1]], ptr [[TMP11]], align 8, !llvm.access.group !4 // CHECK9-NEXT: store ptr [[SIVAR1]], ptr [[TMP11]], align 8, !llvm.access.group [[ACC_GRP4]]
// CHECK9-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(ptr noundef nonnull align 8 dereferenceable(8) [[REF_TMP]]), !llvm.access.group !4 // CHECK9-NEXT: call void @"_ZZZ4mainENK3$_0clEvENKUlvE_clEv"(ptr noundef nonnull align 8 dereferenceable(8) [[REF_TMP]]), !llvm.access.group [[ACC_GRP4]]
// 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: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK9-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP12]], 1 // CHECK9-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK9-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group !4 // CHECK9-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]]
// CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]] // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![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(ptr @[[GLOB1]], i32 [[TMP2]]) // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]])
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK9-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0 // CHECK9-NEXT: [[TMP14:%.*]] = icmp ne i32 [[TMP13]], 0
// CHECK9-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] // CHECK9-NEXT: br i1 [[TMP14]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]]
// CHECK9: .omp.final.then: // CHECK9: .omp.final.then:
// CHECK9-NEXT: store i32 2, ptr [[I]], align 4 // CHECK9-NEXT: store i32 2, ptr [[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: [[TMP15:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK9-NEXT: store ptr [[SIVAR1]], ptr [[TMP15]], align 8 // CHECK9-NEXT: store ptr [[SIVAR1]], ptr [[TMP15]], align 8
// CHECK9-NEXT: [[TMP18:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var) // CHECK9-NEXT: [[TMP16:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP2]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK9-NEXT: switch i32 [[TMP18]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK9-NEXT: switch i32 [[TMP16]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK9-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK9-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK9-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK9-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK9-NEXT: ] // CHECK9-NEXT: ]
// CHECK9: .omp.reduction.case1: // CHECK9: .omp.reduction.case1:
// CHECK9-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK9-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] // CHECK9-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP17]], [[TMP18]]
// CHECK9-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4 // CHECK9-NEXT: store i32 [[ADD5]], ptr [[TMP0]], align 4
// CHECK9-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK9-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK9-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK9-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK9: .omp.reduction.case2: // CHECK9: .omp.reduction.case2:
// CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[SIVAR1]], align 4 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, ptr [[SIVAR1]], align 4
// CHECK9-NEXT: [[TMP22:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP21]] monotonic, align 4 // CHECK9-NEXT: [[TMP20:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP19]] monotonic, align 4
// CHECK9-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var) // CHECK9-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP2]], ptr @.gomp_critical_user_.reduction.var)
// CHECK9-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK9-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK9: .omp.reduction.default: // CHECK9: .omp.reduction.default:
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK9-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK9-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR5:[0-9]+]] { // CHECK9-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR5:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK9-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK9-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK9-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK9-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK9-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK9-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK9-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK9-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK9-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK9-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK9-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK9-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK9-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK9-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
// //
// //
// CHECK9-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK9-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK9-SAME: () #[[ATTR7:[0-9]+]] { // CHECK9-SAME: () #[[ATTR7:[0-9]+]] {
// CHECK9-NEXT: entry: // CHECK9-NEXT: entry:
// CHECK9-NEXT: call void @__tgt_register_requires(i64 1) // CHECK9-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK9-NEXT: ret void // CHECK9-NEXT: ret void
Show All 11 Lines

clang/test/OpenMP/target_teams_map_codegen.cpp

Show First 20 Lines • Show All 80 Lines • ▼ Show 20 Lines
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[Y:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[Y:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP5]], align 8 // CHECK1-NEXT: store ptr [[Y]], ptr [[TMP3]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP7]], align 8 // CHECK1-NEXT: store ptr [[Y]], ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store ptr null, ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP5]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP12]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 2, ptr [[TMP13]], align 4 // CHECK1-NEXT: store i32 2, ptr [[TMP9]], align 4
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP10]], ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP11]], ptr [[TMP15]], align 8 // CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP16]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP17]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP18]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP19]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP15]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 0, ptr [[TMP20]], align 8 // CHECK1-NEXT: store i64 0, ptr [[TMP16]], align 8
// CHECK1-NEXT: [[TMP21:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 // CHECK1-NEXT: [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0
// CHECK1-NEXT: br i1 [[TMP22]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP18]], 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]+}}__Z14mapWithPrivatev_l27() #[[ATTR2:[0-9]+]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27() #[[ATTR2:[0-9]+]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27
Show All 20 Lines
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[Y:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[Y:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP5]], align 8 // CHECK1-NEXT: store ptr [[Y]], ptr [[TMP3]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP7]], align 8 // CHECK1-NEXT: store ptr [[Y]], ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store ptr null, ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP5]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP12]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 2, ptr [[TMP13]], align 4 // CHECK1-NEXT: store i32 2, ptr [[TMP9]], align 4
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP10]], ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP11]], ptr [[TMP15]], align 8 // CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.2, ptr [[TMP16]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.2, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.3, ptr [[TMP17]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.3, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP18]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP19]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP15]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 0, ptr [[TMP20]], align 8 // CHECK1-NEXT: store i64 0, ptr [[TMP16]], align 8
// CHECK1-NEXT: [[TMP21:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 // CHECK1-NEXT: [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0
// CHECK1-NEXT: br i1 [[TMP22]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP18]], 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]+}}__Z19mapWithFirstprivatev_l33(ptr [[X]], ptr [[Y]]) #[[ATTR2]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33(ptr [[X]], ptr [[Y]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33
Show All 36 Lines
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[Y:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[Y:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP5]], align 8 // CHECK1-NEXT: store ptr [[Y]], ptr [[TMP3]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP7]], align 8 // CHECK1-NEXT: store ptr [[Y]], ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store ptr null, ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP5]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP12]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 2, ptr [[TMP13]], align 4 // CHECK1-NEXT: store i32 2, ptr [[TMP9]], align 4
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP10]], ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP11]], ptr [[TMP15]], align 8 // CHECK1-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.5, ptr [[TMP16]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.5, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP17]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP18]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP19]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP15]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 0, ptr [[TMP20]], align 8 // CHECK1-NEXT: store i64 0, ptr [[TMP16]], align 8
// CHECK1-NEXT: [[TMP21:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP17:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 // CHECK1-NEXT: [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0
// CHECK1-NEXT: br i1 [[TMP22]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP18]], 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]+}}__Z16mapWithReductionv_l39(ptr [[X]], ptr [[Y]]) #[[ATTR2]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39(ptr [[X]], ptr [[Y]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39
Show All 24 Lines
// CHECK1-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8
// CHECK1-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8 // CHECK1-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
// CHECK1-NEXT: store i32 0, ptr [[X1]], align 4 // CHECK1-NEXT: store i32 0, ptr [[X1]], align 4
// CHECK1-NEXT: store i32 0, ptr [[Y2]], align 4 // CHECK1-NEXT: store i32 0, ptr [[Y2]], align 4
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[X1]], ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr [[X1]], ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT: store ptr [[Y2]], ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr [[Y2]], ptr [[TMP3]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP7]], i32 2, i64 16, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP5]], i32 2, i64 16, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP9]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[X1]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[X1]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP0]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP0]], align 4
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP1]], align 4 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[Y2]], align 4 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[Y2]], align 4
// CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// CHECK1-NEXT: store i32 [[ADD3]], ptr [[TMP1]], align 4 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[TMP1]], align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP7]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[X1]], align 4 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[X1]], align 4
// CHECK1-NEXT: [[TMP15:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP14]] monotonic, align 4 // CHECK1-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP11]] monotonic, align 4
// CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[Y2]], align 4 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[Y2]], align 4
// CHECK1-NEXT: [[TMP17:%.*]] = atomicrmw add ptr [[TMP1]], i32 [[TMP16]] monotonic, align 4 // CHECK1-NEXT: [[TMP14:%.*]] = atomicrmw add ptr [[TMP1]], i32 [[TMP13]] monotonic, align 4
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP7]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i64 0, i64 1
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP4]], i64 0, i64 1 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 1
// CHECK1-NEXT: [[TMP13:%.*]] = load ptr, ptr [[TMP12]], align 8 // CHECK1-NEXT: [[TMP11:%.*]] = load ptr, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 1 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT: [[TMP16:%.*]] = load ptr, ptr [[TMP15]], align 8 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP11]], align 4
// CHECK1-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP9]], align 4
// CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP16]], align 4 // CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP13]], align 4 // CHECK1-NEXT: store i32 [[ADD2]], ptr [[TMP11]], align 4
// CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
// CHECK1-NEXT: store i32 [[ADD2]], ptr [[TMP16]], align 4
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_Z7mapFromv // CHECK1-LABEL: define {{[^@]+}}@_Z7mapFromv
// CHECK1-SAME: () #[[ATTR0]] { // CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.8, ptr [[TMP11]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.8, ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.9, ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.9, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 0, ptr [[TMP15]], align 8 // CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP15]], 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]+}}__Z7mapFromv_l45(ptr [[X]]) #[[ATTR2]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45(ptr [[X]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45
Show All 26 Lines
// CHECK1-SAME: () #[[ATTR0]] { // CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.11, ptr [[TMP11]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.11, ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.12, ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.12, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 0, ptr [[TMP15]], align 8 // CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP15]], 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]+}}__Z5mapTov_l51(ptr [[X]]) #[[ATTR2]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51(ptr [[X]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51
Show All 26 Lines
// CHECK1-SAME: () #[[ATTR0]] { // CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4 // CHECK1-NEXT: [[X:%.*]] = alloca i32, align 4
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.14, ptr [[TMP11]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.14, ptr [[TMP9]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.15, ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.15, ptr [[TMP10]], align 8
// CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP13]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 0, ptr [[TMP15]], align 8 // CHECK1-NEXT: store i64 0, ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK1-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK1-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP15]], 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]+}}__Z8mapAllocv_l57(ptr [[X]]) #[[ATTR2]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57(ptr [[X]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57
Show All 31 Lines
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [3 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [3 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [3 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP0]], align 8 // CHECK1-NEXT: store ptr [[Y]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr [[Y]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[Z]], ptr [[TMP5]], align 8 // CHECK1-NEXT: store ptr [[Z]], ptr [[TMP3]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[Z]], ptr [[TMP7]], align 8 // CHECK1-NEXT: store ptr [[Z]], ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store ptr null, ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP5]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP17]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP11]], align 4
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 3, ptr [[TMP18]], align 4 // CHECK1-NEXT: store i32 3, ptr [[TMP12]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP19]], align 8 // CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP20]], align 8 // CHECK1-NEXT: store ptr [[TMP10]], ptr [[TMP14]], align 8
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.18, ptr [[TMP21]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.18, ptr [[TMP15]], align 8
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.19, ptr [[TMP22]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.19, ptr [[TMP16]], align 8
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP23]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP17]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP24]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP18]], align 8
// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 0, ptr [[TMP25]], align 8 // CHECK1-NEXT: store i64 0, ptr [[TMP19]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP20:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP27:%.*]] = icmp ne i32 [[TMP26]], 0 // CHECK1-NEXT: [[TMP21:%.*]] = icmp ne i32 [[TMP20]], 0
// CHECK1-NEXT: br i1 [[TMP27]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP21]], 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]+}}__Z8mapArrayv_l63(ptr [[Y]], ptr [[Z]]) #[[ATTR2]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63(ptr [[Y]], ptr [[Z]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP28]], align 8 // CHECK1-NEXT: store ptr [[Y]], ptr [[TMP22]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP30]], align 8 // CHECK1-NEXT: store ptr [[Y]], ptr [[TMP23]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP32]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP24]], align 8
// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 1 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[Z]], ptr [[TMP33]], align 8 // CHECK1-NEXT: store ptr [[Z]], ptr [[TMP25]], align 8
// CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 1 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[Z]], ptr [[TMP35]], align 8 // CHECK1-NEXT: store ptr [[Z]], ptr [[TMP26]], align 8
// CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 1 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 1
// CHECK1-NEXT: store ptr null, ptr [[TMP37]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP27]], align 8
// CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 2 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP38]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP28]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 2 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP40]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP29]], align 8
// CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 2 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 2
// CHECK1-NEXT: store ptr null, ptr [[TMP42]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP30]], align 8
// CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS4:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS4:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 0 // CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP45]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP33]], align 4
// CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 1 // CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 1
// CHECK1-NEXT: store i32 3, ptr [[TMP46]], align 4 // CHECK1-NEXT: store i32 3, ptr [[TMP34]], align 4
// CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 2 // CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP43]], ptr [[TMP47]], align 8 // CHECK1-NEXT: store ptr [[TMP31]], ptr [[TMP35]], align 8
// CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 3 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP44]], ptr [[TMP48]], align 8 // CHECK1-NEXT: store ptr [[TMP32]], ptr [[TMP36]], align 8
// CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 4 // CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.22, ptr [[TMP49]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.22, ptr [[TMP37]], align 8
// CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 5 // CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.23, ptr [[TMP50]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.23, ptr [[TMP38]], align 8
// CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 6 // CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP51]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP39]], align 8
// CHECK1-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 7 // CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP52]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP40]], align 8
// CHECK1-NEXT: [[TMP53:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 8 // CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 8
// CHECK1-NEXT: store i64 0, ptr [[TMP53]], align 8 // CHECK1-NEXT: store i64 0, ptr [[TMP41]], align 8
// CHECK1-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65.region_id, ptr [[KERNEL_ARGS4]]) // CHECK1-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65.region_id, ptr [[KERNEL_ARGS4]])
// CHECK1-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0 // CHECK1-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK1-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK1-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK1: omp_offload.failed5: // CHECK1: omp_offload.failed5:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65(ptr [[Y]], ptr [[Z]]) #[[ATTR2]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65(ptr [[Y]], ptr [[Z]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK1: omp_offload.cont6: // CHECK1: omp_offload.cont6:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63
Show All 23 Lines
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8 // CHECK1-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
// CHECK1-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8 // CHECK1-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i64 352, i1 false) // CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i64 352, i1 false)
// CHECK1-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0 // CHECK1-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i64 99 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i64 99
// CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP4]] // CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1: omp.arrayinit.body: // CHECK1: omp.arrayinit.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK1-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1: omp.arrayinit.done: // CHECK1: omp.arrayinit.done:
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[Z2]], ptr [[TMP5]], align 8 // CHECK1-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP8]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.17, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.17, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP10]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP11]] // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP7]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1: omp.arraycpy.body: // CHECK1: omp.arraycpy.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP11]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP7]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1: omp.arraycpy.done6: // CHECK1: omp.arraycpy.done6:
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP8]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP14]] // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP10]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]]
// CHECK1: omp.arraycpy.body8: // CHECK1: omp.arraycpy.body8:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4
// CHECK1-NEXT: [[TMP16:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP15]] monotonic, align 4 // CHECK1-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP11]] monotonic, align 4
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP14]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP10]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]]
// CHECK1: omp.arraycpy.done14: // CHECK1: omp.arraycpy.done14:
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP8]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.17 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.17
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i64 99
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[TMP10]], i64 99
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP10]], [[TMP12]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1: omp.arraycpy.body: // CHECK1: omp.arraycpy.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP10]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1: omp.arraycpy.done2: // CHECK1: omp.arraycpy.done2:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
Show All 21 Lines
// CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8 // CHECK1-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
// CHECK1-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8 // CHECK1-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
// CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i64 352, i1 false) // CHECK1-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i64 352, i1 false)
// CHECK1-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0 // CHECK1-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i64 99 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i64 99
// CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP4]] // CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1: omp.arrayinit.body: // CHECK1: omp.arrayinit.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK1-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1: omp.arrayinit.done: // CHECK1: omp.arrayinit.done:
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[Z2]], ptr [[TMP5]], align 8 // CHECK1-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK1-NEXT: [[TMP10:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP8]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.21, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.21, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP10]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP11:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP11]] // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP7]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1: omp.arraycpy.body: // CHECK1: omp.arraycpy.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP11]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP7]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1: omp.arraycpy.done6: // CHECK1: omp.arraycpy.done6:
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP8]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP14]] // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP10]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]]
// CHECK1: omp.arraycpy.body8: // CHECK1: omp.arraycpy.body8:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4
// CHECK1-NEXT: [[TMP16:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP15]] monotonic, align 4 // CHECK1-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP11]] monotonic, align 4
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP14]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP10]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]]
// CHECK1: omp.arraycpy.done14: // CHECK1: omp.arraycpy.done14:
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP8]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.21 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.21
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i64 99
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[TMP10]], i64 99
// CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP10]], [[TMP12]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1: omp.arraycpy.body: // CHECK1: omp.arraycpy.body:
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP10]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// CHECK1-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK1-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]] // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]] // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1: omp.arraycpy.done2: // CHECK1: omp.arraycpy.done2:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@_Z9mapInt128v // CHECK1-LABEL: define {{[^@]+}}@_Z9mapInt128v
// CHECK1-SAME: () #[[ATTR0]] { // CHECK1-SAME: () #[[ATTR0]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[X:%.*]] = alloca i128, align 16 // CHECK1-NEXT: [[X:%.*]] = alloca i128, align 16
// CHECK1-NEXT: [[Y:%.*]] = alloca i128, align 16 // CHECK1-NEXT: [[Y:%.*]] = alloca i128, align 16
// CHECK1-NEXT: [[Z:%.*]] = alloca i128, align 16 // CHECK1-NEXT: [[Z:%.*]] = alloca i128, align 16
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [3 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [3 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [3 x ptr], align 8 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP0]], align 8 // CHECK1-NEXT: store ptr [[Y]], ptr [[TMP0]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP2]], align 8 // CHECK1-NEXT: store ptr [[Y]], ptr [[TMP1]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP4]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[Z]], ptr [[TMP5]], align 8 // CHECK1-NEXT: store ptr [[Z]], ptr [[TMP3]], align 8
// CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[Z]], ptr [[TMP7]], align 8 // CHECK1-NEXT: store ptr [[Z]], ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1
// CHECK1-NEXT: store ptr null, ptr [[TMP9]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP5]], align 8
// CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP10]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP12]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP7]], align 8
// CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2
// CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP8]], align 8
// CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP17]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP11]], align 4
// CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK1-NEXT: store i32 3, ptr [[TMP18]], align 4 // CHECK1-NEXT: store i32 3, ptr [[TMP12]], align 4
// CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP15]], ptr [[TMP19]], align 8 // CHECK1-NEXT: store ptr [[TMP9]], ptr [[TMP13]], align 8
// CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP16]], ptr [[TMP20]], align 8 // CHECK1-NEXT: store ptr [[TMP10]], ptr [[TMP14]], align 8
// CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.26, ptr [[TMP21]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.26, ptr [[TMP15]], align 8
// CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.27, ptr [[TMP22]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.27, ptr [[TMP16]], align 8
// CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP23]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP17]], align 8
// CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP24]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP18]], align 8
// CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK1-NEXT: store i64 0, ptr [[TMP25]], align 8 // CHECK1-NEXT: store i64 0, ptr [[TMP19]], align 8
// CHECK1-NEXT: [[TMP26:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l72.region_id, ptr [[KERNEL_ARGS]]) // CHECK1-NEXT: [[TMP20:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l72.region_id, ptr [[KERNEL_ARGS]])
// CHECK1-NEXT: [[TMP27:%.*]] = icmp ne i32 [[TMP26]], 0 // CHECK1-NEXT: [[TMP21:%.*]] = icmp ne i32 [[TMP20]], 0
// CHECK1-NEXT: br i1 [[TMP27]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK1-NEXT: br i1 [[TMP21]], 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]+}}__Z9mapInt128v_l72(ptr [[Y]], ptr [[Z]]) #[[ATTR2]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l72(ptr [[Y]], ptr [[Z]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK1: omp_offload.cont: // CHECK1: omp_offload.cont:
// CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP28]], align 8 // CHECK1-NEXT: store ptr [[Y]], ptr [[TMP22]], align 8
// CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK1-NEXT: store ptr [[Y]], ptr [[TMP30]], align 8 // CHECK1-NEXT: store ptr [[Y]], ptr [[TMP23]], align 8
// CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0
// CHECK1-NEXT: store ptr null, ptr [[TMP32]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP24]], align 8
// CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 1 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[Z]], ptr [[TMP33]], align 8 // CHECK1-NEXT: store ptr [[Z]], ptr [[TMP25]], align 8
// CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 1 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 1
// CHECK1-NEXT: store ptr [[Z]], ptr [[TMP35]], align 8 // CHECK1-NEXT: store ptr [[Z]], ptr [[TMP26]], align 8
// CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 1 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 1
// CHECK1-NEXT: store ptr null, ptr [[TMP37]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP27]], align 8
// CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 2 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP38]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP28]], align 8
// CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 2 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[X]], ptr [[TMP40]], align 8 // CHECK1-NEXT: store ptr [[X]], ptr [[TMP29]], align 8
// CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 2 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 2
// CHECK1-NEXT: store ptr null, ptr [[TMP42]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP30]], align 8
// CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK1-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK1-NEXT: [[KERNEL_ARGS4:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK1-NEXT: [[KERNEL_ARGS4:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 0 // CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 0
// CHECK1-NEXT: store i32 1, ptr [[TMP45]], align 4 // CHECK1-NEXT: store i32 1, ptr [[TMP33]], align 4
// CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 1 // CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 1
// CHECK1-NEXT: store i32 3, ptr [[TMP46]], align 4 // CHECK1-NEXT: store i32 3, ptr [[TMP34]], align 4
// CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 2 // CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 2
// CHECK1-NEXT: store ptr [[TMP43]], ptr [[TMP47]], align 8 // CHECK1-NEXT: store ptr [[TMP31]], ptr [[TMP35]], align 8
// CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 3 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 3
// CHECK1-NEXT: store ptr [[TMP44]], ptr [[TMP48]], align 8 // CHECK1-NEXT: store ptr [[TMP32]], ptr [[TMP36]], align 8
// CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 4 // CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 4
// CHECK1-NEXT: store ptr @.offload_sizes.30, ptr [[TMP49]], align 8 // CHECK1-NEXT: store ptr @.offload_sizes.30, ptr [[TMP37]], align 8
// CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 5 // CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 5
// CHECK1-NEXT: store ptr @.offload_maptypes.31, ptr [[TMP50]], align 8 // CHECK1-NEXT: store ptr @.offload_maptypes.31, ptr [[TMP38]], align 8
// CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 6 // CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 6
// CHECK1-NEXT: store ptr null, ptr [[TMP51]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP39]], align 8
// CHECK1-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 7 // CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 7
// CHECK1-NEXT: store ptr null, ptr [[TMP52]], align 8 // CHECK1-NEXT: store ptr null, ptr [[TMP40]], align 8
// CHECK1-NEXT: [[TMP53:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 8 // CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 8
// CHECK1-NEXT: store i64 0, ptr [[TMP53]], align 8 // CHECK1-NEXT: store i64 0, ptr [[TMP41]], align 8
// CHECK1-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l74.region_id, ptr [[KERNEL_ARGS4]]) // CHECK1-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l74.region_id, ptr [[KERNEL_ARGS4]])
// CHECK1-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0 // CHECK1-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK1-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK1-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK1: omp_offload.failed5: // CHECK1: omp_offload.failed5:
// CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l74(ptr [[Y]], ptr [[Z]]) #[[ATTR2]] // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l74(ptr [[Y]], ptr [[Z]]) #[[ATTR2]]
// CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK1: omp_offload.cont6: // CHECK1: omp_offload.cont6:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l72 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l72
Show All 29 Lines
// CHECK1-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8 // CHECK1-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i128, ptr [[TMP0]], align 16 // CHECK1-NEXT: [[TMP2:%.*]] = load i128, ptr [[TMP0]], align 16
// CHECK1-NEXT: store i128 [[TMP2]], ptr [[Y1]], align 16 // CHECK1-NEXT: store i128 [[TMP2]], ptr [[Y1]], align 16
// CHECK1-NEXT: store i128 0, ptr [[Z2]], align 16 // CHECK1-NEXT: store i128 0, ptr [[Z2]], align 16
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8 // CHECK1-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP6]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.25, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.25, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP8]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP9:%.*]] = load i128, ptr [[TMP1]], align 16 // CHECK1-NEXT: [[TMP7:%.*]] = load i128, ptr [[TMP1]], align 16
// CHECK1-NEXT: [[TMP10:%.*]] = load i128, ptr [[Z2]], align 16 // CHECK1-NEXT: [[TMP8:%.*]] = load i128, ptr [[Z2]], align 16
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP9]], [[TMP10]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP7]], [[TMP8]]
// CHECK1-NEXT: store i128 [[ADD]], ptr [[TMP1]], align 16 // CHECK1-NEXT: store i128 [[ADD]], ptr [[TMP1]], align 16
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP6]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP11:%.*]] = load i128, ptr [[Z2]], align 16 // CHECK1-NEXT: [[TMP9:%.*]] = load i128, ptr [[Z2]], align 16
// CHECK1-NEXT: call void @__atomic_load(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], i32 noundef signext 0) // CHECK1-NEXT: call void @__atomic_load(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], i32 noundef signext 0)
// CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]] // CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]]
// CHECK1: atomic_cont: // CHECK1: atomic_cont:
// CHECK1-NEXT: [[TMP14:%.*]] = load i128, ptr [[ATOMIC_TEMP]], align 16 // CHECK1-NEXT: [[TMP10:%.*]] = load i128, ptr [[ATOMIC_TEMP]], align 16
// CHECK1-NEXT: store i128 [[TMP14]], ptr [[TMP]], align 16 // CHECK1-NEXT: store i128 [[TMP10]], ptr [[TMP]], align 16
// CHECK1-NEXT: [[TMP15:%.*]] = load i128, ptr [[TMP]], align 16 // CHECK1-NEXT: [[TMP11:%.*]] = load i128, ptr [[TMP]], align 16
// CHECK1-NEXT: [[TMP16:%.*]] = load i128, ptr [[Z2]], align 16 // CHECK1-NEXT: [[TMP12:%.*]] = load i128, ptr [[Z2]], align 16
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i128 [[TMP15]], [[TMP16]] // CHECK1-NEXT: [[ADD4:%.*]] = add nsw i128 [[TMP11]], [[TMP12]]
// CHECK1-NEXT: store i128 [[ADD4]], ptr [[ATOMIC_TEMP3]], align 16 // CHECK1-NEXT: store i128 [[ADD4]], ptr [[ATOMIC_TEMP3]], align 16
// CHECK1-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @__atomic_compare_exchange(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], ptr noundef [[ATOMIC_TEMP3]], i32 noundef signext 0, i32 noundef signext 0) // CHECK1-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @__atomic_compare_exchange(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], ptr noundef [[ATOMIC_TEMP3]], i32 noundef signext 0, i32 noundef signext 0)
// CHECK1-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]] // CHECK1-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]]
// CHECK1: atomic_exit: // CHECK1: atomic_exit:
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP6]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.25 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.25
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = load i128, ptr [[TMP7]], align 16
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load i128, ptr [[TMP5]], align 16
// CHECK1-NEXT: [[TMP12:%.*]] = load i128, ptr [[TMP10]], align 16 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i128, ptr [[TMP7]], align 16 // CHECK1-NEXT: store i128 [[ADD]], ptr [[TMP7]], align 16
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i128 [[ADD]], ptr [[TMP10]], align 16
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l74 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l74
// CHECK1-SAME: (ptr noundef nonnull align 16 dereferenceable(16) [[Y:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Z:%.*]]) #[[ATTR1]] { // CHECK1-SAME: (ptr noundef nonnull align 16 dereferenceable(16) [[Y:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Z:%.*]]) #[[ATTR1]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
Show All 25 Lines
// CHECK1-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8 // CHECK1-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
// CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
// CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load i128, ptr [[TMP0]], align 16 // CHECK1-NEXT: [[TMP2:%.*]] = load i128, ptr [[TMP0]], align 16
// CHECK1-NEXT: store i128 [[TMP2]], ptr [[Y1]], align 16 // CHECK1-NEXT: store i128 [[TMP2]], ptr [[Y1]], align 16
// CHECK1-NEXT: store i128 0, ptr [[Z2]], align 16 // CHECK1-NEXT: store i128 0, ptr [[Z2]], align 16
// CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8 // CHECK1-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK1-NEXT: [[TMP8:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP6]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.29, ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.29, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: switch i32 [[TMP8]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK1-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT: ] // CHECK1-NEXT: ]
// CHECK1: .omp.reduction.case1: // CHECK1: .omp.reduction.case1:
// CHECK1-NEXT: [[TMP9:%.*]] = load i128, ptr [[TMP1]], align 16 // CHECK1-NEXT: [[TMP7:%.*]] = load i128, ptr [[TMP1]], align 16
// CHECK1-NEXT: [[TMP10:%.*]] = load i128, ptr [[Z2]], align 16 // CHECK1-NEXT: [[TMP8:%.*]] = load i128, ptr [[Z2]], align 16
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP9]], [[TMP10]] // CHECK1-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP7]], [[TMP8]]
// CHECK1-NEXT: store i128 [[ADD]], ptr [[TMP1]], align 16 // CHECK1-NEXT: store i128 [[ADD]], ptr [[TMP1]], align 16
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP6]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.case2: // CHECK1: .omp.reduction.case2:
// CHECK1-NEXT: [[TMP11:%.*]] = load i128, ptr [[Z2]], align 16 // CHECK1-NEXT: [[TMP9:%.*]] = load i128, ptr [[Z2]], align 16
// CHECK1-NEXT: call void @__atomic_load(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], i32 noundef signext 0) // CHECK1-NEXT: call void @__atomic_load(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], i32 noundef signext 0)
// CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]] // CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]]
// CHECK1: atomic_cont: // CHECK1: atomic_cont:
// CHECK1-NEXT: [[TMP14:%.*]] = load i128, ptr [[ATOMIC_TEMP]], align 16 // CHECK1-NEXT: [[TMP10:%.*]] = load i128, ptr [[ATOMIC_TEMP]], align 16
// CHECK1-NEXT: store i128 [[TMP14]], ptr [[TMP]], align 16 // CHECK1-NEXT: store i128 [[TMP10]], ptr [[TMP]], align 16
// CHECK1-NEXT: [[TMP15:%.*]] = load i128, ptr [[TMP]], align 16 // CHECK1-NEXT: [[TMP11:%.*]] = load i128, ptr [[TMP]], align 16
// CHECK1-NEXT: [[TMP16:%.*]] = load i128, ptr [[Z2]], align 16 // CHECK1-NEXT: [[TMP12:%.*]] = load i128, ptr [[Z2]], align 16
// CHECK1-NEXT: [[ADD4:%.*]] = add nsw i128 [[TMP15]], [[TMP16]] // CHECK1-NEXT: [[ADD4:%.*]] = add nsw i128 [[TMP11]], [[TMP12]]
// CHECK1-NEXT: store i128 [[ADD4]], ptr [[ATOMIC_TEMP3]], align 16 // CHECK1-NEXT: store i128 [[ADD4]], ptr [[ATOMIC_TEMP3]], align 16
// CHECK1-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @__atomic_compare_exchange(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], ptr noundef [[ATOMIC_TEMP3]], i32 noundef signext 0, i32 noundef signext 0) // CHECK1-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @__atomic_compare_exchange(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], ptr noundef [[ATOMIC_TEMP3]], i32 noundef signext 0, i32 noundef signext 0)
// CHECK1-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]] // CHECK1-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]]
// CHECK1: atomic_exit: // CHECK1: atomic_exit:
// CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP6]], ptr @.gomp_critical_user_.reduction.var) // CHECK1-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1: .omp.reduction.default: // CHECK1: .omp.reduction.default:
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.29 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.29
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK1-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK1-NEXT: [[TMP8:%.*]] = load i128, ptr [[TMP7]], align 16
// CHECK1-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK1-NEXT: [[TMP9:%.*]] = load i128, ptr [[TMP5]], align 16
// CHECK1-NEXT: [[TMP12:%.*]] = load i128, ptr [[TMP10]], align 16 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP8]], [[TMP9]]
// CHECK1-NEXT: [[TMP13:%.*]] = load i128, ptr [[TMP7]], align 16 // CHECK1-NEXT: store i128 [[ADD]], ptr [[TMP7]], align 16
// CHECK1-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP12]], [[TMP13]]
// CHECK1-NEXT: store i128 [[ADD]], ptr [[TMP10]], align 16
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK1-SAME: () #[[ATTR7:[0-9]+]] { // CHECK1-SAME: () #[[ATTR7:[0-9]+]] {
// CHECK1-NEXT: entry: // CHECK1-NEXT: entry:
// CHECK1-NEXT: call void @__tgt_register_requires(i64 1) // CHECK1-NEXT: call void @__tgt_register_requires(i64 1)
// CHECK1-NEXT: ret void // CHECK1-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@_Z14mapWithPrivatev // CHECK3-LABEL: define {{[^@]+}}@_Z14mapWithPrivatev
// CHECK3-SAME: () #[[ATTR0:[0-9]+]] { // CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[Y:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[Y:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP2]], align 4 // CHECK3-NEXT: store ptr [[X]], ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP5]], align 4 // CHECK3-NEXT: store ptr [[Y]], ptr [[TMP3]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP7]], align 4 // CHECK3-NEXT: store ptr [[Y]], ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK3-NEXT: store ptr null, ptr [[TMP9]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP12]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 2, ptr [[TMP13]], align 4 // CHECK3-NEXT: store i32 2, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP10]], ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP11]], ptr [[TMP15]], align 4 // CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP16]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP17]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP13]], align 4
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP18]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP19]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP15]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 0, ptr [[TMP20]], align 8 // CHECK3-NEXT: store i64 0, ptr [[TMP16]], align 8
// CHECK3-NEXT: [[TMP21:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP17:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 // CHECK3-NEXT: [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0
// CHECK3-NEXT: br i1 [[TMP22]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP18]], 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]+}}__Z14mapWithPrivatev_l27() #[[ATTR2:[0-9]+]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27() #[[ATTR2:[0-9]+]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27
Show All 20 Lines
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[Y:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[Y:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP2]], align 4 // CHECK3-NEXT: store ptr [[X]], ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP5]], align 4 // CHECK3-NEXT: store ptr [[Y]], ptr [[TMP3]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP7]], align 4 // CHECK3-NEXT: store ptr [[Y]], ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK3-NEXT: store ptr null, ptr [[TMP9]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP12]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 2, ptr [[TMP13]], align 4 // CHECK3-NEXT: store i32 2, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP10]], ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP11]], ptr [[TMP15]], align 4 // CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes.2, ptr [[TMP16]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes.2, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes.3, ptr [[TMP17]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes.3, ptr [[TMP13]], align 4
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP18]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP19]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP15]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 0, ptr [[TMP20]], align 8 // CHECK3-NEXT: store i64 0, ptr [[TMP16]], align 8
// CHECK3-NEXT: [[TMP21:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP17:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 // CHECK3-NEXT: [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0
// CHECK3-NEXT: br i1 [[TMP22]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP18]], 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]+}}__Z19mapWithFirstprivatev_l33(ptr [[X]], ptr [[Y]]) #[[ATTR2]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33(ptr [[X]], ptr [[Y]]) #[[ATTR2]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z19mapWithFirstprivatev_l33
Show All 36 Lines
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[Y:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[Y:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [2 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [2 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [2 x ptr], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP2]], align 4 // CHECK3-NEXT: store ptr [[X]], ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP5]], align 4 // CHECK3-NEXT: store ptr [[Y]], ptr [[TMP3]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP7]], align 4 // CHECK3-NEXT: store ptr [[Y]], ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK3-NEXT: store ptr null, ptr [[TMP9]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP12]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 2, ptr [[TMP13]], align 4 // CHECK3-NEXT: store i32 2, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP10]], ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP11]], ptr [[TMP15]], align 4 // CHECK3-NEXT: store ptr [[TMP7]], ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes.5, ptr [[TMP16]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes.5, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP17]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP13]], align 4
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP18]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP19]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP15]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 0, ptr [[TMP20]], align 8 // CHECK3-NEXT: store i64 0, ptr [[TMP16]], align 8
// CHECK3-NEXT: [[TMP21:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP17:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 // CHECK3-NEXT: [[TMP18:%.*]] = icmp ne i32 [[TMP17]], 0
// CHECK3-NEXT: br i1 [[TMP22]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP18]], 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]+}}__Z16mapWithReductionv_l39(ptr [[X]], ptr [[Y]]) #[[ATTR2]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39(ptr [[X]], ptr [[Y]]) #[[ATTR2]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z16mapWithReductionv_l39
Show All 24 Lines
// CHECK3-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 4 // CHECK3-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 4
// CHECK3-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4 // CHECK3-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
// CHECK3-NEXT: store i32 0, ptr [[X1]], align 4 // CHECK3-NEXT: store i32 0, ptr [[X1]], align 4
// CHECK3-NEXT: store i32 0, ptr [[Y2]], align 4 // CHECK3-NEXT: store i32 0, ptr [[Y2]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[X1]], ptr [[TMP2]], align 4 // CHECK3-NEXT: store ptr [[X1]], ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 1 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 1
// CHECK3-NEXT: store ptr [[Y2]], ptr [[TMP4]], align 4 // CHECK3-NEXT: store ptr [[Y2]], ptr [[TMP3]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP7]], i32 2, i32 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP5]], i32 2, i32 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: switch i32 [[TMP9]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK3-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK3-NEXT: ] // CHECK3-NEXT: ]
// CHECK3: .omp.reduction.case1: // CHECK3: .omp.reduction.case1:
// CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[X1]], align 4 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[X1]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP1]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[Y2]], align 4 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[Y2]], align 4
// CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// CHECK3-NEXT: store i32 [[ADD3]], ptr [[TMP1]], align 4 // CHECK3-NEXT: store i32 [[ADD3]], ptr [[TMP1]], align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP7]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.case2: // CHECK3: .omp.reduction.case2:
// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[X1]], align 4 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[X1]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP14]] monotonic, align 4 // CHECK3-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP11]] monotonic, align 4
// CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[Y2]], align 4 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[Y2]], align 4
// CHECK3-NEXT: [[TMP17:%.*]] = atomicrmw add ptr [[TMP1]], i32 [[TMP16]] monotonic, align 4 // CHECK3-NEXT: [[TMP14:%.*]] = atomicrmw add ptr [[TMP1]], i32 [[TMP13]] monotonic, align 4
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP7]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.default: // CHECK3: .omp.reduction.default:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { // CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i32 0, i32 1
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP4]], i32 0, i32 1 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i32 0, i32 1
// CHECK3-NEXT: [[TMP13:%.*]] = load ptr, ptr [[TMP12]], align 4 // CHECK3-NEXT: [[TMP11:%.*]] = load ptr, ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i32 0, i32 1 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = load ptr, ptr [[TMP15]], align 4 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP11]], align 4
// CHECK3-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP16]], align 4 // CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK3-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP13]], align 4 // CHECK3-NEXT: store i32 [[ADD2]], ptr [[TMP11]], align 4
// CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
// CHECK3-NEXT: store i32 [[ADD2]], ptr [[TMP16]], align 4
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@_Z7mapFromv // CHECK3-LABEL: define {{[^@]+}}@_Z7mapFromv
// CHECK3-SAME: () #[[ATTR0]] { // CHECK3-SAME: () #[[ATTR0]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP2]], align 4 // CHECK3-NEXT: store ptr [[X]], ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes.8, ptr [[TMP11]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes.8, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes.9, ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes.9, ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 0, ptr [[TMP15]], align 8 // CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP15]], 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]+}}__Z7mapFromv_l45(ptr [[X]]) #[[ATTR2]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45(ptr [[X]]) #[[ATTR2]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45
Show All 26 Lines
// CHECK3-SAME: () #[[ATTR0]] { // CHECK3-SAME: () #[[ATTR0]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP2]], align 4 // CHECK3-NEXT: store ptr [[X]], ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes.11, ptr [[TMP11]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes.11, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes.12, ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes.12, ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 0, ptr [[TMP15]], align 8 // CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP15]], 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]+}}__Z5mapTov_l51(ptr [[X]]) #[[ATTR2]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51(ptr [[X]]) #[[ATTR2]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5mapTov_l51
Show All 26 Lines
// CHECK3-SAME: () #[[ATTR0]] { // CHECK3-SAME: () #[[ATTR0]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4 // CHECK3-NEXT: [[X:%.*]] = alloca i32, align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store ptr [[X]], ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP2]], align 4 // CHECK3-NEXT: store ptr [[X]], ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP7]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 1, ptr [[TMP8]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP5]], ptr [[TMP9]], align 4 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP6]], ptr [[TMP10]], align 4 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes.14, ptr [[TMP11]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes.14, ptr [[TMP9]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes.15, ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes.15, ptr [[TMP10]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP13]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 0, ptr [[TMP15]], align 8 // CHECK3-NEXT: store i64 0, ptr [[TMP13]], align 8
// CHECK3-NEXT: [[TMP16:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP14:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP17:%.*]] = icmp ne i32 [[TMP16]], 0 // CHECK3-NEXT: [[TMP15:%.*]] = icmp ne i32 [[TMP14]], 0
// CHECK3-NEXT: br i1 [[TMP17]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP15]], 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]+}}__Z8mapAllocv_l57(ptr [[X]]) #[[ATTR2]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57(ptr [[X]]) #[[ATTR2]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapAllocv_l57
Show All 31 Lines
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [3 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [3 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [3 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [3 x ptr], align 4
// CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [3 x ptr], align 4 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [3 x ptr], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP0]], align 4 // CHECK3-NEXT: store ptr [[Y]], ptr [[TMP0]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP2]], align 4 // CHECK3-NEXT: store ptr [[Y]], ptr [[TMP1]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP4]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1
// CHECK3-NEXT: store ptr [[Z]], ptr [[TMP5]], align 4 // CHECK3-NEXT: store ptr [[Z]], ptr [[TMP3]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1
// CHECK3-NEXT: store ptr [[Z]], ptr [[TMP7]], align 4 // CHECK3-NEXT: store ptr [[Z]], ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1
// CHECK3-NEXT: store ptr null, ptr [[TMP9]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP5]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP10]], align 4 // CHECK3-NEXT: store ptr [[X]], ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP12]], align 4 // CHECK3-NEXT: store ptr [[X]], ptr [[TMP7]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2
// CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP8]], align 4
// CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0
// CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8
// CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP17]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP11]], align 4
// CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1
// CHECK3-NEXT: store i32 3, ptr [[TMP18]], align 4 // CHECK3-NEXT: store i32 3, ptr [[TMP12]], align 4
// CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP15]], ptr [[TMP19]], align 4 // CHECK3-NEXT: store ptr [[TMP9]], ptr [[TMP13]], align 4
// CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP16]], ptr [[TMP20]], align 4 // CHECK3-NEXT: store ptr [[TMP10]], ptr [[TMP14]], align 4
// CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes.18, ptr [[TMP21]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes.18, ptr [[TMP15]], align 4
// CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes.19, ptr [[TMP22]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes.19, ptr [[TMP16]], align 4
// CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP23]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP17]], align 4
// CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 // CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP24]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP18]], align 4
// CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 // CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8
// CHECK3-NEXT: store i64 0, ptr [[TMP25]], align 8 // CHECK3-NEXT: store i64 0, ptr [[TMP19]], align 8
// CHECK3-NEXT: [[TMP26:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63.region_id, ptr [[KERNEL_ARGS]]) // CHECK3-NEXT: [[TMP20:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63.region_id, ptr [[KERNEL_ARGS]])
// CHECK3-NEXT: [[TMP27:%.*]] = icmp ne i32 [[TMP26]], 0 // CHECK3-NEXT: [[TMP21:%.*]] = icmp ne i32 [[TMP20]], 0
// CHECK3-NEXT: br i1 [[TMP27]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] // CHECK3-NEXT: br i1 [[TMP21]], 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]+}}__Z8mapArrayv_l63(ptr [[Y]], ptr [[Z]]) #[[ATTR2]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63(ptr [[Y]], ptr [[Z]]) #[[ATTR2]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]]
// CHECK3: omp_offload.cont: // CHECK3: omp_offload.cont:
// CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP28]], align 4 // CHECK3-NEXT: store ptr [[Y]], ptr [[TMP22]], align 4
// CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[Y]], ptr [[TMP30]], align 4 // CHECK3-NEXT: store ptr [[Y]], ptr [[TMP23]], align 4
// CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0
// CHECK3-NEXT: store ptr null, ptr [[TMP32]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP24]], align 4
// CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 1 // CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 1
// CHECK3-NEXT: store ptr [[Z]], ptr [[TMP33]], align 4 // CHECK3-NEXT: store ptr [[Z]], ptr [[TMP25]], align 4
// CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 1 // CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 1
// CHECK3-NEXT: store ptr [[Z]], ptr [[TMP35]], align 4 // CHECK3-NEXT: store ptr [[Z]], ptr [[TMP26]], align 4
// CHECK3-NEXT: [[TMP37:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 1 // CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 1
// CHECK3-NEXT: store ptr null, ptr [[TMP37]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP27]], align 4
// CHECK3-NEXT: [[TMP38:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 2 // CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP38]], align 4 // CHECK3-NEXT: store ptr [[X]], ptr [[TMP28]], align 4
// CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 2 // CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[X]], ptr [[TMP40]], align 4 // CHECK3-NEXT: store ptr [[X]], ptr [[TMP29]], align 4
// CHECK3-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 2 // CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 2
// CHECK3-NEXT: store ptr null, ptr [[TMP42]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP30]], align 4
// CHECK3-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 // CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0
// CHECK3-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 // CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0
// CHECK3-NEXT: [[KERNEL_ARGS4:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 // CHECK3-NEXT: [[KERNEL_ARGS4:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8
// CHECK3-NEXT: [[TMP45:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 0 // CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 0
// CHECK3-NEXT: store i32 1, ptr [[TMP45]], align 4 // CHECK3-NEXT: store i32 1, ptr [[TMP33]], align 4
// CHECK3-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 1 // CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 1
// CHECK3-NEXT: store i32 3, ptr [[TMP46]], align 4 // CHECK3-NEXT: store i32 3, ptr [[TMP34]], align 4
// CHECK3-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 2 // CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 2
// CHECK3-NEXT: store ptr [[TMP43]], ptr [[TMP47]], align 4 // CHECK3-NEXT: store ptr [[TMP31]], ptr [[TMP35]], align 4
// CHECK3-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 3 // CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 3
// CHECK3-NEXT: store ptr [[TMP44]], ptr [[TMP48]], align 4 // CHECK3-NEXT: store ptr [[TMP32]], ptr [[TMP36]], align 4
// CHECK3-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 4 // CHECK3-NEXT: [[TMP37:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 4
// CHECK3-NEXT: store ptr @.offload_sizes.22, ptr [[TMP49]], align 4 // CHECK3-NEXT: store ptr @.offload_sizes.22, ptr [[TMP37]], align 4
// CHECK3-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 5 // CHECK3-NEXT: [[TMP38:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 5
// CHECK3-NEXT: store ptr @.offload_maptypes.23, ptr [[TMP50]], align 4 // CHECK3-NEXT: store ptr @.offload_maptypes.23, ptr [[TMP38]], align 4
// CHECK3-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 6 // CHECK3-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 6
// CHECK3-NEXT: store ptr null, ptr [[TMP51]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP39]], align 4
// CHECK3-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 7 // CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 7
// CHECK3-NEXT: store ptr null, ptr [[TMP52]], align 4 // CHECK3-NEXT: store ptr null, ptr [[TMP40]], align 4
// CHECK3-NEXT: [[TMP53:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 8 // CHECK3-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS4]], i32 0, i32 8
// CHECK3-NEXT: store i64 0, ptr [[TMP53]], align 8 // CHECK3-NEXT: store i64 0, ptr [[TMP41]], align 8
// CHECK3-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65.region_id, ptr [[KERNEL_ARGS4]]) // CHECK3-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB1]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65.region_id, ptr [[KERNEL_ARGS4]])
// CHECK3-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0 // CHECK3-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0
// CHECK3-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]] // CHECK3-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED5:%.*]], label [[OMP_OFFLOAD_CONT6:%.*]]
// CHECK3: omp_offload.failed5: // CHECK3: omp_offload.failed5:
// CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65(ptr [[Y]], ptr [[Z]]) #[[ATTR2]] // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65(ptr [[Y]], ptr [[Z]]) #[[ATTR2]]
// CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT6]] // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT6]]
// CHECK3: omp_offload.cont6: // CHECK3: omp_offload.cont6:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l63
Show All 23 Lines
// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4 // CHECK3-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
// CHECK3-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 4 // CHECK3-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 4
// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i32 352, i1 false) // CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i32 352, i1 false)
// CHECK3-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0 // CHECK3-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i32 99 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i32 99
// CHECK3-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP4]] // CHECK3-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
// CHECK3-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] // CHECK3-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK3: omp.arrayinit.body: // CHECK3: omp.arrayinit.body:
// CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] // CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK3-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK3-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK3-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]] // CHECK3-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK3: omp.arrayinit.done: // CHECK3: omp.arrayinit.done:
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[Z2]], ptr [[TMP5]], align 4 // CHECK3-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP8]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.17, ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.17, ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: switch i32 [[TMP10]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK3-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK3-NEXT: ] // CHECK3-NEXT: ]
// CHECK3: .omp.reduction.case1: // CHECK3: .omp.reduction.case1:
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99
// CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP11]] // CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP7]]
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK3: omp.arraycpy.body: // CHECK3: omp.arraycpy.body:
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1 // CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK3-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP11]] // CHECK3-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP7]]
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]] // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
// CHECK3: omp.arraycpy.done6: // CHECK3: omp.arraycpy.done6:
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP8]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.case2: // CHECK3: .omp.reduction.case2:
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99
// CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP14]] // CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP10]]
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]] // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]]
// CHECK3: omp.arraycpy.body8: // CHECK3: omp.arraycpy.body8:
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ] // CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ] // CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP15]] monotonic, align 4 // CHECK3-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP11]] monotonic, align 4
// CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1 // CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1 // CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1
// CHECK3-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP14]] // CHECK3-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP10]]
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]] // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]]
// CHECK3: omp.arraycpy.done14: // CHECK3: omp.arraycpy.done14:
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP8]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.default: // CHECK3: .omp.reduction.default:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.17 // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.17
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i32 99
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[TMP10]], i32 99
// CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP10]], [[TMP12]]
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK3: omp.arraycpy.body: // CHECK3: omp.arraycpy.body:
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP10]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK3-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]] // CHECK3-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]] // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK3: omp.arraycpy.done2: // CHECK3: omp.arraycpy.done2:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65
// CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR1]] { // CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR1]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
Show All 21 Lines
// CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK3-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4 // CHECK3-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
// CHECK3-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 4 // CHECK3-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 4
// CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 4 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
// CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 4 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 4
// CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i32 352, i1 false) // CHECK3-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i32 352, i1 false)
// CHECK3-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0 // CHECK3-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP4:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i32 99 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i32 99
// CHECK3-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP4]] // CHECK3-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
// CHECK3-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] // CHECK3-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK3: omp.arrayinit.body: // CHECK3: omp.arrayinit.body:
// CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] // CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK3-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK3-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK3-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]] // CHECK3-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK3: omp.arrayinit.done: // CHECK3: omp.arrayinit.done:
// CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK3-NEXT: store ptr [[Z2]], ptr [[TMP5]], align 4 // CHECK3-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 4
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP10:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP8]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.21, ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.21, ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: switch i32 [[TMP10]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK3-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK3-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK3-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK3-NEXT: ] // CHECK3-NEXT: ]
// CHECK3: .omp.reduction.case1: // CHECK3: .omp.reduction.case1:
// CHECK3-NEXT: [[TMP11:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99
// CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP11]] // CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP7]]
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK3: omp.arraycpy.body: // CHECK3: omp.arraycpy.body:
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1 // CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK3-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP11]] // CHECK3-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP7]]
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]] // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
// CHECK3: omp.arraycpy.done6: // CHECK3: omp.arraycpy.done6:
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP8]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.case2: // CHECK3: .omp.reduction.case2:
// CHECK3-NEXT: [[TMP14:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99
// CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP14]] // CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP10]]
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]] // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]]
// CHECK3: omp.arraycpy.body8: // CHECK3: omp.arraycpy.body8:
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ] // CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ] // CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4
// CHECK3-NEXT: [[TMP16:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP15]] monotonic, align 4 // CHECK3-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP11]] monotonic, align 4
// CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1 // CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1 // CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1
// CHECK3-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP14]] // CHECK3-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP10]]
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]] // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]]
// CHECK3: omp.arraycpy.done14: // CHECK3: omp.arraycpy.done14:
// CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP8]], ptr @.gomp_critical_user_.reduction.var) // CHECK3-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK3-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK3: .omp.reduction.default: // CHECK3: .omp.reduction.default:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.21 // CHECK3-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.21
// CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] { // CHECK3-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR3]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
// CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK3-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK3-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK3-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK3-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i32 99
// CHECK3-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// CHECK3-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[TMP10]], i32 99
// CHECK3-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP10]], [[TMP12]]
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK3: omp.arraycpy.body: // CHECK3: omp.arraycpy.body:
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK3-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP10]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK3-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]] // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// CHECK3-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK3-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK3-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK3-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK3-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]] // CHECK3-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]] // CHECK3-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK3: omp.arraycpy.done2: // CHECK3: omp.arraycpy.done2:
// CHECK3-NEXT: ret void // CHECK3-NEXT: ret void
// //
// //
// CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg // CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg
// CHECK3-SAME: () #[[ATTR6:[0-9]+]] { // CHECK3-SAME: () #[[ATTR6:[0-9]+]] {
// CHECK3-NEXT: entry: // CHECK3-NEXT: entry:
▲ Show 20 Lines • Show All 83 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8 // CHECK5-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 8
// CHECK5-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8 // CHECK5-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 8
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 8 // CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
// CHECK5-NEXT: store i32 0, ptr [[X1]], align 4 // CHECK5-NEXT: store i32 0, ptr [[X1]], align 4
// CHECK5-NEXT: store i32 0, ptr [[Y2]], align 4 // CHECK5-NEXT: store i32 0, ptr [[Y2]], align 4
// CHECK5-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK5-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK5-NEXT: store ptr [[X1]], ptr [[TMP2]], align 8 // CHECK5-NEXT: store ptr [[X1]], ptr [[TMP2]], align 8
// CHECK5-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 // CHECK5-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK5-NEXT: store ptr [[Y2]], ptr [[TMP4]], align 8 // CHECK5-NEXT: store ptr [[Y2]], ptr [[TMP3]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK5-NEXT: [[TMP9:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP7]], i32 2, i64 16, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP5]], i32 2, i64 16, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: switch i32 [[TMP9]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK5-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK5-NEXT: ] // CHECK5-NEXT: ]
// CHECK5: .omp.reduction.case1: // CHECK5: .omp.reduction.case1:
// CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[X1]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[X1]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[TMP0]], align 4 // CHECK5-NEXT: store i32 [[ADD]], ptr [[TMP0]], align 4
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP1]], align 4 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[Y2]], align 4 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[Y2]], align 4
// CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK5-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// CHECK5-NEXT: store i32 [[ADD3]], ptr [[TMP1]], align 4 // CHECK5-NEXT: store i32 [[ADD3]], ptr [[TMP1]], align 4
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP7]], ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.case2: // CHECK5: .omp.reduction.case2:
// CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[X1]], align 4 // CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[X1]], align 4
// CHECK5-NEXT: [[TMP15:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP14]] monotonic, align 4 // CHECK5-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP11]] monotonic, align 4
// CHECK5-NEXT: [[TMP16:%.*]] = load i32, ptr [[Y2]], align 4 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[Y2]], align 4
// CHECK5-NEXT: [[TMP17:%.*]] = atomicrmw add ptr [[TMP1]], i32 [[TMP16]] monotonic, align 4 // CHECK5-NEXT: [[TMP14:%.*]] = atomicrmw add ptr [[TMP1]], i32 [[TMP13]] monotonic, align 4
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP7]], ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.default: // CHECK5: .omp.reduction.default:
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK5-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK5-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK5-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i64 0, i64 1
// CHECK5-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK5-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
// CHECK5-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP4]], i64 0, i64 1 // CHECK5-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 1
// CHECK5-NEXT: [[TMP13:%.*]] = load ptr, ptr [[TMP12]], align 8 // CHECK5-NEXT: [[TMP11:%.*]] = load ptr, ptr [[TMP10]], align 8
// CHECK5-NEXT: [[TMP15:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 1 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK5-NEXT: [[TMP16:%.*]] = load ptr, ptr [[TMP15]], align 8 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK5-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK5-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK5-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP11]], align 4
// CHECK5-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4 // CHECK5-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP9]], align 4
// CHECK5-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP16]], align 4 // CHECK5-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK5-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP13]], align 4 // CHECK5-NEXT: store i32 [[ADD2]], ptr [[TMP11]], align 4
// CHECK5-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
// CHECK5-NEXT: store i32 [[ADD2]], ptr [[TMP16]], align 4
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45
// CHECK5-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]]) #[[ATTR0]] { // CHECK5-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]]) #[[ATTR0]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: [[X_CASTED:%.*]] = alloca i64, align 8 // CHECK5-NEXT: [[X_CASTED:%.*]] = alloca i64, align 8
▲ Show 20 Lines • Show All 97 Lines • ▼ Show 20 Lines
// CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK5-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8 // CHECK5-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
// CHECK5-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8 // CHECK5-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8 // CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
// CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i64 352, i1 false) // CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i64 352, i1 false)
// CHECK5-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0 // CHECK5-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0
// CHECK5-NEXT: [[TMP4:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i64 99 // CHECK5-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i64 99
// CHECK5-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP4]] // CHECK5-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
// CHECK5-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] // CHECK5-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK5: omp.arrayinit.body: // CHECK5: omp.arrayinit.body:
// CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] // CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK5-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK5-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK5-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]] // CHECK5-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] // CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK5: omp.arrayinit.done: // CHECK5: omp.arrayinit.done:
// CHECK5-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK5-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK5-NEXT: store ptr [[Z2]], ptr [[TMP5]], align 8 // CHECK5-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8
// CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK5-NEXT: [[TMP10:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP8]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.7, ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.7, ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: switch i32 [[TMP10]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK5-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK5-NEXT: ] // CHECK5-NEXT: ]
// CHECK5: .omp.reduction.case1: // CHECK5: .omp.reduction.case1:
// CHECK5-NEXT: [[TMP11:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99 // CHECK5-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99
// CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP11]] // CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP7]]
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK5: omp.arraycpy.body: // CHECK5: omp.arraycpy.body:
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4 // CHECK5-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1 // CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK5-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP11]] // CHECK5-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP7]]
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]] // CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
// CHECK5: omp.arraycpy.done6: // CHECK5: omp.arraycpy.done6:
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP8]], ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.case2: // CHECK5: .omp.reduction.case2:
// CHECK5-NEXT: [[TMP14:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99 // CHECK5-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99
// CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP14]] // CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP10]]
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]] // CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]]
// CHECK5: omp.arraycpy.body8: // CHECK5: omp.arraycpy.body8:
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ] // CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ] // CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK5-NEXT: [[TMP15:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4 // CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4
// CHECK5-NEXT: [[TMP16:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP15]] monotonic, align 4 // CHECK5-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP11]] monotonic, align 4
// CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1 // CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1 // CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1
// CHECK5-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP14]] // CHECK5-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP10]]
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]] // CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]]
// CHECK5: omp.arraycpy.done14: // CHECK5: omp.arraycpy.done14:
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP8]], ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.default: // CHECK5: .omp.reduction.default:
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.7 // CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.7
// CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] { // CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK5-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK5-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK5-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i64 99
// CHECK5-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// CHECK5-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[TMP10]], i64 99
// CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP10]], [[TMP12]]
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK5: omp.arraycpy.body: // CHECK5: omp.arraycpy.body:
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP10]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]] // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK5-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK5-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]] // CHECK5-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]] // CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK5: omp.arraycpy.done2: // CHECK5: omp.arraycpy.done2:
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65
// CHECK5-SAME: (ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR0]] { // CHECK5-SAME: (ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR0]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
Show All 21 Lines
// CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 // CHECK5-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK5-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8 // CHECK5-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 8
// CHECK5-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8 // CHECK5-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8 // CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
// CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i64 352, i1 false) // CHECK5-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i64 352, i1 false)
// CHECK5-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0 // CHECK5-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0
// CHECK5-NEXT: [[TMP4:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i64 99 // CHECK5-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i64 99
// CHECK5-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP4]] // CHECK5-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
// CHECK5-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] // CHECK5-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK5: omp.arrayinit.body: // CHECK5: omp.arrayinit.body:
// CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] // CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK5-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK5-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK5-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]] // CHECK5-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] // CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK5: omp.arrayinit.done: // CHECK5: omp.arrayinit.done:
// CHECK5-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK5-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK5-NEXT: store ptr [[Z2]], ptr [[TMP5]], align 8 // CHECK5-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8
// CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK5-NEXT: [[TMP10:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP8]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.9, ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.9, ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: switch i32 [[TMP10]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK5-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK5-NEXT: ] // CHECK5-NEXT: ]
// CHECK5: .omp.reduction.case1: // CHECK5: .omp.reduction.case1:
// CHECK5-NEXT: [[TMP11:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99 // CHECK5-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99
// CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP11]] // CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP7]]
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK5: omp.arraycpy.body: // CHECK5: omp.arraycpy.body:
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4 // CHECK5-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1 // CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK5-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP11]] // CHECK5-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP7]]
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]] // CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
// CHECK5: omp.arraycpy.done6: // CHECK5: omp.arraycpy.done6:
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP8]], ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.case2: // CHECK5: .omp.reduction.case2:
// CHECK5-NEXT: [[TMP14:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99 // CHECK5-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[TMP1]], i64 99
// CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP14]] // CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP10]]
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]] // CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]]
// CHECK5: omp.arraycpy.body8: // CHECK5: omp.arraycpy.body8:
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ] // CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ] // CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK5-NEXT: [[TMP15:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4 // CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4
// CHECK5-NEXT: [[TMP16:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP15]] monotonic, align 4 // CHECK5-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP11]] monotonic, align 4
// CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1 // CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1 // CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1
// CHECK5-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP14]] // CHECK5-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP10]]
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]] // CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]]
// CHECK5: omp.arraycpy.done14: // CHECK5: omp.arraycpy.done14:
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP8]], ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.default: // CHECK5: .omp.reduction.default:
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.9 // CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.9
// CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] { // CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK5-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK5-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK5-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i64 99
// CHECK5-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// CHECK5-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[TMP10]], i64 99
// CHECK5-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP10]], [[TMP12]]
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK5: omp.arraycpy.body: // CHECK5: omp.arraycpy.body:
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK5-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP10]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK5-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]] // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// CHECK5-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK5-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK5-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK5-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK5-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]] // CHECK5-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]] // CHECK5-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK5: omp.arraycpy.done2: // CHECK5: omp.arraycpy.done2:
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l72 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l72
// CHECK5-SAME: (ptr noundef nonnull align 16 dereferenceable(16) [[Y:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Z:%.*]]) #[[ATTR0]] { // CHECK5-SAME: (ptr noundef nonnull align 16 dereferenceable(16) [[Y:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Z:%.*]]) #[[ATTR0]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
Show All 27 Lines
// CHECK5-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8 // CHECK5-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8 // CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
// CHECK5-NEXT: [[TMP2:%.*]] = load i128, ptr [[TMP0]], align 16 // CHECK5-NEXT: [[TMP2:%.*]] = load i128, ptr [[TMP0]], align 16
// CHECK5-NEXT: store i128 [[TMP2]], ptr [[Y1]], align 16 // CHECK5-NEXT: store i128 [[TMP2]], ptr [[Y1]], align 16
// CHECK5-NEXT: store i128 0, ptr [[Z2]], align 16 // CHECK5-NEXT: store i128 0, ptr [[Z2]], align 16
// CHECK5-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK5-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK5-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8 // CHECK5-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8
// CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK5-NEXT: [[TMP8:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP6]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.11, ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.11, ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: switch i32 [[TMP8]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK5-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK5-NEXT: ] // CHECK5-NEXT: ]
// CHECK5: .omp.reduction.case1: // CHECK5: .omp.reduction.case1:
// CHECK5-NEXT: [[TMP9:%.*]] = load i128, ptr [[TMP1]], align 16 // CHECK5-NEXT: [[TMP7:%.*]] = load i128, ptr [[TMP1]], align 16
// CHECK5-NEXT: [[TMP10:%.*]] = load i128, ptr [[Z2]], align 16 // CHECK5-NEXT: [[TMP8:%.*]] = load i128, ptr [[Z2]], align 16
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP9]], [[TMP10]] // CHECK5-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP7]], [[TMP8]]
// CHECK5-NEXT: store i128 [[ADD]], ptr [[TMP1]], align 16 // CHECK5-NEXT: store i128 [[ADD]], ptr [[TMP1]], align 16
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP6]], ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.case2: // CHECK5: .omp.reduction.case2:
// CHECK5-NEXT: [[TMP11:%.*]] = load i128, ptr [[Z2]], align 16 // CHECK5-NEXT: [[TMP9:%.*]] = load i128, ptr [[Z2]], align 16
// CHECK5-NEXT: call void @__atomic_load(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], i32 noundef signext 0) #[[ATTR6:[0-9]+]] // CHECK5-NEXT: call void @__atomic_load(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], i32 noundef signext 0) #[[ATTR6:[0-9]+]]
// CHECK5-NEXT: br label [[ATOMIC_CONT:%.*]] // CHECK5-NEXT: br label [[ATOMIC_CONT:%.*]]
// CHECK5: atomic_cont: // CHECK5: atomic_cont:
// CHECK5-NEXT: [[TMP14:%.*]] = load i128, ptr [[ATOMIC_TEMP]], align 16 // CHECK5-NEXT: [[TMP10:%.*]] = load i128, ptr [[ATOMIC_TEMP]], align 16
// CHECK5-NEXT: store i128 [[TMP14]], ptr [[TMP]], align 16 // CHECK5-NEXT: store i128 [[TMP10]], ptr [[TMP]], align 16
// CHECK5-NEXT: [[TMP15:%.*]] = load i128, ptr [[TMP]], align 16 // CHECK5-NEXT: [[TMP11:%.*]] = load i128, ptr [[TMP]], align 16
// CHECK5-NEXT: [[TMP16:%.*]] = load i128, ptr [[Z2]], align 16 // CHECK5-NEXT: [[TMP12:%.*]] = load i128, ptr [[Z2]], align 16
// CHECK5-NEXT: [[ADD4:%.*]] = add nsw i128 [[TMP15]], [[TMP16]] // CHECK5-NEXT: [[ADD4:%.*]] = add nsw i128 [[TMP11]], [[TMP12]]
// CHECK5-NEXT: store i128 [[ADD4]], ptr [[ATOMIC_TEMP3]], align 16 // CHECK5-NEXT: store i128 [[ADD4]], ptr [[ATOMIC_TEMP3]], align 16
// CHECK5-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @__atomic_compare_exchange(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], ptr noundef [[ATOMIC_TEMP3]], i32 noundef signext 0, i32 noundef signext 0) #[[ATTR6]] // CHECK5-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @__atomic_compare_exchange(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], ptr noundef [[ATOMIC_TEMP3]], i32 noundef signext 0, i32 noundef signext 0) #[[ATTR6]]
// CHECK5-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]] // CHECK5-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]]
// CHECK5: atomic_exit: // CHECK5: atomic_exit:
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP6]], ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.default: // CHECK5: .omp.reduction.default:
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.11 // CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.11
// CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] { // CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK5-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK5-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK5-NEXT: [[TMP8:%.*]] = load i128, ptr [[TMP7]], align 16
// CHECK5-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK5-NEXT: [[TMP9:%.*]] = load i128, ptr [[TMP5]], align 16
// CHECK5-NEXT: [[TMP12:%.*]] = load i128, ptr [[TMP10]], align 16 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP8]], [[TMP9]]
// CHECK5-NEXT: [[TMP13:%.*]] = load i128, ptr [[TMP7]], align 16 // CHECK5-NEXT: store i128 [[ADD]], ptr [[TMP7]], align 16
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP12]], [[TMP13]]
// CHECK5-NEXT: store i128 [[ADD]], ptr [[TMP10]], align 16
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l74 // CHECK5-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z9mapInt128v_l74
// CHECK5-SAME: (ptr noundef nonnull align 16 dereferenceable(16) [[Y:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Z:%.*]]) #[[ATTR0]] { // CHECK5-SAME: (ptr noundef nonnull align 16 dereferenceable(16) [[Y:%.*]], ptr noundef nonnull align 16 dereferenceable(16) [[Z:%.*]]) #[[ATTR0]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[Y_ADDR:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[Z_ADDR:%.*]] = alloca ptr, align 8
Show All 25 Lines
// CHECK5-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8 // CHECK5-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 8
// CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8 // CHECK5-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 8
// CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8 // CHECK5-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 8
// CHECK5-NEXT: [[TMP2:%.*]] = load i128, ptr [[TMP0]], align 16 // CHECK5-NEXT: [[TMP2:%.*]] = load i128, ptr [[TMP0]], align 16
// CHECK5-NEXT: store i128 [[TMP2]], ptr [[Y1]], align 16 // CHECK5-NEXT: store i128 [[TMP2]], ptr [[Y1]], align 16
// CHECK5-NEXT: store i128 0, ptr [[Z2]], align 16 // CHECK5-NEXT: store i128 0, ptr [[Z2]], align 16
// CHECK5-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 // CHECK5-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK5-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8 // CHECK5-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 8
// CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 // CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK5-NEXT: [[TMP8:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP6]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.13, ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.13, ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: switch i32 [[TMP8]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK5-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK5-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK5-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK5-NEXT: ] // CHECK5-NEXT: ]
// CHECK5: .omp.reduction.case1: // CHECK5: .omp.reduction.case1:
// CHECK5-NEXT: [[TMP9:%.*]] = load i128, ptr [[TMP1]], align 16 // CHECK5-NEXT: [[TMP7:%.*]] = load i128, ptr [[TMP1]], align 16
// CHECK5-NEXT: [[TMP10:%.*]] = load i128, ptr [[Z2]], align 16 // CHECK5-NEXT: [[TMP8:%.*]] = load i128, ptr [[Z2]], align 16
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP9]], [[TMP10]] // CHECK5-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP7]], [[TMP8]]
// CHECK5-NEXT: store i128 [[ADD]], ptr [[TMP1]], align 16 // CHECK5-NEXT: store i128 [[ADD]], ptr [[TMP1]], align 16
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP6]], ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.case2: // CHECK5: .omp.reduction.case2:
// CHECK5-NEXT: [[TMP11:%.*]] = load i128, ptr [[Z2]], align 16 // CHECK5-NEXT: [[TMP9:%.*]] = load i128, ptr [[Z2]], align 16
// CHECK5-NEXT: call void @__atomic_load(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], i32 noundef signext 0) #[[ATTR6]] // CHECK5-NEXT: call void @__atomic_load(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], i32 noundef signext 0) #[[ATTR6]]
// CHECK5-NEXT: br label [[ATOMIC_CONT:%.*]] // CHECK5-NEXT: br label [[ATOMIC_CONT:%.*]]
// CHECK5: atomic_cont: // CHECK5: atomic_cont:
// CHECK5-NEXT: [[TMP14:%.*]] = load i128, ptr [[ATOMIC_TEMP]], align 16 // CHECK5-NEXT: [[TMP10:%.*]] = load i128, ptr [[ATOMIC_TEMP]], align 16
// CHECK5-NEXT: store i128 [[TMP14]], ptr [[TMP]], align 16 // CHECK5-NEXT: store i128 [[TMP10]], ptr [[TMP]], align 16
// CHECK5-NEXT: [[TMP15:%.*]] = load i128, ptr [[TMP]], align 16 // CHECK5-NEXT: [[TMP11:%.*]] = load i128, ptr [[TMP]], align 16
// CHECK5-NEXT: [[TMP16:%.*]] = load i128, ptr [[Z2]], align 16 // CHECK5-NEXT: [[TMP12:%.*]] = load i128, ptr [[Z2]], align 16
// CHECK5-NEXT: [[ADD4:%.*]] = add nsw i128 [[TMP15]], [[TMP16]] // CHECK5-NEXT: [[ADD4:%.*]] = add nsw i128 [[TMP11]], [[TMP12]]
// CHECK5-NEXT: store i128 [[ADD4]], ptr [[ATOMIC_TEMP3]], align 16 // CHECK5-NEXT: store i128 [[ADD4]], ptr [[ATOMIC_TEMP3]], align 16
// CHECK5-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @__atomic_compare_exchange(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], ptr noundef [[ATOMIC_TEMP3]], i32 noundef signext 0, i32 noundef signext 0) #[[ATTR6]] // CHECK5-NEXT: [[CALL:%.*]] = call noundef zeroext i1 @__atomic_compare_exchange(i64 noundef 16, ptr noundef [[TMP1]], ptr noundef [[ATOMIC_TEMP]], ptr noundef [[ATOMIC_TEMP3]], i32 noundef signext 0, i32 noundef signext 0) #[[ATTR6]]
// CHECK5-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]] // CHECK5-NEXT: br i1 [[CALL]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]]
// CHECK5: atomic_exit: // CHECK5: atomic_exit:
// CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP6]], ptr @.gomp_critical_user_.reduction.var) // CHECK5-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK5-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK5: .omp.reduction.default: // CHECK5: .omp.reduction.default:
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.13 // CHECK5-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.13
// CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] { // CHECK5-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] {
// CHECK5-NEXT: entry: // CHECK5-NEXT: entry:
// CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8 // CHECK5-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8 // CHECK5-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8 // CHECK5-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK5-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 8 // CHECK5-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i64 0, i64 0 // CHECK5-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK5-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK5-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8 // CHECK5-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK5-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0 // CHECK5-NEXT: [[TMP8:%.*]] = load i128, ptr [[TMP7]], align 16
// CHECK5-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 8 // CHECK5-NEXT: [[TMP9:%.*]] = load i128, ptr [[TMP5]], align 16
// CHECK5-NEXT: [[TMP12:%.*]] = load i128, ptr [[TMP10]], align 16 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP8]], [[TMP9]]
// CHECK5-NEXT: [[TMP13:%.*]] = load i128, ptr [[TMP7]], align 16 // CHECK5-NEXT: store i128 [[ADD]], ptr [[TMP7]], align 16
// CHECK5-NEXT: [[ADD:%.*]] = add nsw i128 [[TMP12]], [[TMP13]]
// CHECK5-NEXT: store i128 [[ADD]], ptr [[TMP10]], align 16
// CHECK5-NEXT: ret void // CHECK5-NEXT: ret void
// //
// //
// CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z14mapWithPrivatev_l27
// CHECK7-SAME: () #[[ATTR0:[0-9]+]] { // CHECK7-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK7-NEXT: entry: // CHECK7-NEXT: entry:
// CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1:[0-9]+]], i32 0, ptr @.omp_outlined.) // CHECK7-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB1:[0-9]+]], i32 0, ptr @.omp_outlined.)
// CHECK7-NEXT: ret void // CHECK7-NEXT: ret void
▲ Show 20 Lines • Show All 74 Lines • ▼ Show 20 Lines
// CHECK7-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 4 // CHECK7-NEXT: store ptr [[X]], ptr [[X_ADDR]], align 4
// CHECK7-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4 // CHECK7-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
// CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 4 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[X_ADDR]], align 4
// CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 4 // CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
// CHECK7-NEXT: store i32 0, ptr [[X1]], align 4 // CHECK7-NEXT: store i32 0, ptr [[X1]], align 4
// CHECK7-NEXT: store i32 0, ptr [[Y2]], align 4 // CHECK7-NEXT: store i32 0, ptr [[Y2]], align 4
// CHECK7-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK7-NEXT: [[TMP2:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK7-NEXT: store ptr [[X1]], ptr [[TMP2]], align 4 // CHECK7-NEXT: store ptr [[X1]], ptr [[TMP2]], align 4
// CHECK7-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 1 // CHECK7-NEXT: [[TMP3:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 1
// CHECK7-NEXT: store ptr [[Y2]], ptr [[TMP4]], align 4 // CHECK7-NEXT: store ptr [[Y2]], ptr [[TMP3]], align 4
// CHECK7-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK7-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK7-NEXT: [[TMP9:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP7]], i32 2, i32 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var) // CHECK7-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP5]], i32 2, i32 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK7-NEXT: switch i32 [[TMP9]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK7-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK7-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK7-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK7-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK7-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK7-NEXT: ] // CHECK7-NEXT: ]
// CHECK7: .omp.reduction.case1: // CHECK7: .omp.reduction.case1:
// CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP0]], align 4 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[X1]], align 4 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[X1]], align 4
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP7]], [[TMP8]]
// CHECK7-NEXT: store i32 [[ADD]], ptr [[TMP0]], align 4 // CHECK7-NEXT: store i32 [[ADD]], ptr [[TMP0]], align 4
// CHECK7-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP1]], align 4 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP1]], align 4
// CHECK7-NEXT: [[TMP13:%.*]] = load i32, ptr [[Y2]], align 4 // CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[Y2]], align 4
// CHECK7-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK7-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// CHECK7-NEXT: store i32 [[ADD3]], ptr [[TMP1]], align 4 // CHECK7-NEXT: store i32 [[ADD3]], ptr [[TMP1]], align 4
// CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP7]], ptr @.gomp_critical_user_.reduction.var) // CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK7: .omp.reduction.case2: // CHECK7: .omp.reduction.case2:
// CHECK7-NEXT: [[TMP14:%.*]] = load i32, ptr [[X1]], align 4 // CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[X1]], align 4
// CHECK7-NEXT: [[TMP15:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP14]] monotonic, align 4 // CHECK7-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP11]] monotonic, align 4
// CHECK7-NEXT: [[TMP16:%.*]] = load i32, ptr [[Y2]], align 4 // CHECK7-NEXT: [[TMP13:%.*]] = load i32, ptr [[Y2]], align 4
// CHECK7-NEXT: [[TMP17:%.*]] = atomicrmw add ptr [[TMP1]], i32 [[TMP16]] monotonic, align 4 // CHECK7-NEXT: [[TMP14:%.*]] = atomicrmw add ptr [[TMP1]], i32 [[TMP13]] monotonic, align 4
// CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP7]], ptr @.gomp_critical_user_.reduction.var) // CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK7: .omp.reduction.default: // CHECK7: .omp.reduction.default:
// CHECK7-NEXT: ret void // CHECK7-NEXT: ret void
// //
// //
// CHECK7-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func // CHECK7-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func
// CHECK7-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2:[0-9]+]] { // CHECK7-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK7-NEXT: entry: // CHECK7-NEXT: entry:
// CHECK7-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK7-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK7-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK7-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK7-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK7-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK7-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK7-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK7-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK7-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK7-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK7-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK7-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK7-NEXT: [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK7-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK7-NEXT: [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK7-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK7-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK7-NEXT: [[TMP9:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK7-NEXT: [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i32 0, i32 1
// CHECK7-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK7-NEXT: [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 4
// CHECK7-NEXT: [[TMP12:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP4]], i32 0, i32 1 // CHECK7-NEXT: [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i32 0, i32 1
// CHECK7-NEXT: [[TMP13:%.*]] = load ptr, ptr [[TMP12]], align 4 // CHECK7-NEXT: [[TMP11:%.*]] = load ptr, ptr [[TMP10]], align 4
// CHECK7-NEXT: [[TMP15:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i32 0, i32 1 // CHECK7-NEXT: [[TMP12:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK7-NEXT: [[TMP16:%.*]] = load ptr, ptr [[TMP15]], align 4 // CHECK7-NEXT: [[TMP13:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK7-NEXT: [[TMP18:%.*]] = load i32, ptr [[TMP10]], align 4 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]]
// CHECK7-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK7-NEXT: store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] // CHECK7-NEXT: [[TMP14:%.*]] = load i32, ptr [[TMP11]], align 4
// CHECK7-NEXT: store i32 [[ADD]], ptr [[TMP10]], align 4 // CHECK7-NEXT: [[TMP15:%.*]] = load i32, ptr [[TMP9]], align 4
// CHECK7-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP16]], align 4 // CHECK7-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]]
// CHECK7-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP13]], align 4 // CHECK7-NEXT: store i32 [[ADD2]], ptr [[TMP11]], align 4
// CHECK7-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
// CHECK7-NEXT: store i32 [[ADD2]], ptr [[TMP16]], align 4
// CHECK7-NEXT: ret void // CHECK7-NEXT: ret void
// //
// //
// CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z7mapFromv_l45
// CHECK7-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]]) #[[ATTR0]] { // CHECK7-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[X:%.*]]) #[[ATTR0]] {
// CHECK7-NEXT: entry: // CHECK7-NEXT: entry:
// CHECK7-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 4 // CHECK7-NEXT: [[X_ADDR:%.*]] = alloca ptr, align 4
// CHECK7-NEXT: [[X_CASTED:%.*]] = alloca i32, align 4 // CHECK7-NEXT: [[X_CASTED:%.*]] = alloca i32, align 4
▲ Show 20 Lines • Show All 97 Lines • ▼ Show 20 Lines
// CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 // CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK7-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4 // CHECK7-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
// CHECK7-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 4 // CHECK7-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 4
// CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 4 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
// CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 4 // CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 4
// CHECK7-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i32 352, i1 false) // CHECK7-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i32 352, i1 false)
// CHECK7-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0 // CHECK7-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0
// CHECK7-NEXT: [[TMP4:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i32 99 // CHECK7-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i32 99
// CHECK7-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP4]] // CHECK7-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
// CHECK7-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] // CHECK7-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK7: omp.arrayinit.body: // CHECK7: omp.arrayinit.body:
// CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] // CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK7-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK7-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK7-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]] // CHECK7-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] // CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK7: omp.arrayinit.done: // CHECK7: omp.arrayinit.done:
// CHECK7-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK7-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK7-NEXT: store ptr [[Z2]], ptr [[TMP5]], align 4 // CHECK7-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 4
// CHECK7-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK7-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK7-NEXT: [[TMP10:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP8]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.7, ptr @.gomp_critical_user_.reduction.var) // CHECK7-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.7, ptr @.gomp_critical_user_.reduction.var)
// CHECK7-NEXT: switch i32 [[TMP10]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK7-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK7-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK7-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK7-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK7-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK7-NEXT: ] // CHECK7-NEXT: ]
// CHECK7: .omp.reduction.case1: // CHECK7: .omp.reduction.case1:
// CHECK7-NEXT: [[TMP11:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99 // CHECK7-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99
// CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP11]] // CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP7]]
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK7: omp.arraycpy.body: // CHECK7: omp.arraycpy.body:
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK7-NEXT: [[TMP12:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK7-NEXT: [[TMP13:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK7-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4 // CHECK7-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1 // CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK7-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP11]] // CHECK7-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP7]]
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]] // CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
// CHECK7: omp.arraycpy.done6: // CHECK7: omp.arraycpy.done6:
// CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP8]], ptr @.gomp_critical_user_.reduction.var) // CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK7: .omp.reduction.case2: // CHECK7: .omp.reduction.case2:
// CHECK7-NEXT: [[TMP14:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99 // CHECK7-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99
// CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP14]] // CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP10]]
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]] // CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]]
// CHECK7: omp.arraycpy.body8: // CHECK7: omp.arraycpy.body8:
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ] // CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ] // CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK7-NEXT: [[TMP15:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4 // CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4
// CHECK7-NEXT: [[TMP16:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP15]] monotonic, align 4 // CHECK7-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP11]] monotonic, align 4
// CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1 // CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1 // CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1
// CHECK7-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP14]] // CHECK7-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP10]]
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]] // CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]]
// CHECK7: omp.arraycpy.done14: // CHECK7: omp.arraycpy.done14:
// CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP8]], ptr @.gomp_critical_user_.reduction.var) // CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK7: .omp.reduction.default: // CHECK7: .omp.reduction.default:
// CHECK7-NEXT: ret void // CHECK7-NEXT: ret void
// //
// //
// CHECK7-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.7 // CHECK7-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.7
// CHECK7-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] { // CHECK7-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] {
// CHECK7-NEXT: entry: // CHECK7-NEXT: entry:
// CHECK7-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK7-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK7-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK7-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK7-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK7-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK7-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK7-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK7-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK7-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK7-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK7-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK7-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK7-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK7-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK7-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK7-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK7-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK7-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK7-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i32 99
// CHECK7-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// CHECK7-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[TMP10]], i32 99
// CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP10]], [[TMP12]]
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK7: omp.arraycpy.body: // CHECK7: omp.arraycpy.body:
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP10]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK7-NEXT: [[TMP13:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK7-NEXT: [[TMP14:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4 // CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]] // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// CHECK7-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK7-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK7-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]] // CHECK7-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]] // CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK7: omp.arraycpy.done2: // CHECK7: omp.arraycpy.done2:
// CHECK7-NEXT: ret void // CHECK7-NEXT: ret void
// //
// //
// CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65 // CHECK7-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z8mapArrayv_l65
// CHECK7-SAME: (ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR0]] { // CHECK7-SAME: (ptr noundef nonnull align 4 dereferenceable(352) [[Y:%.*]], ptr noundef nonnull align 4 dereferenceable(396) [[Z:%.*]]) #[[ATTR0]] {
// CHECK7-NEXT: entry: // CHECK7-NEXT: entry:
Show All 21 Lines
// CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK7-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 // CHECK7-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4
// CHECK7-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4 // CHECK7-NEXT: store ptr [[Y]], ptr [[Y_ADDR]], align 4
// CHECK7-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 4 // CHECK7-NEXT: store ptr [[Z]], ptr [[Z_ADDR]], align 4
// CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 4 // CHECK7-NEXT: [[TMP0:%.*]] = load ptr, ptr [[Y_ADDR]], align 4
// CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 4 // CHECK7-NEXT: [[TMP1:%.*]] = load ptr, ptr [[Z_ADDR]], align 4
// CHECK7-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i32 352, i1 false) // CHECK7-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[Y1]], ptr align 4 [[TMP0]], i32 352, i1 false)
// CHECK7-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0 // CHECK7-NEXT: [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [99 x i32], ptr [[Z2]], i32 0, i32 0
// CHECK7-NEXT: [[TMP4:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i32 99 // CHECK7-NEXT: [[TMP2:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i32 99
// CHECK7-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP4]] // CHECK7-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP2]]
// CHECK7-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] // CHECK7-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK7: omp.arrayinit.body: // CHECK7: omp.arrayinit.body:
// CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] // CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK7-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK7-NEXT: store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK7-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]] // CHECK7-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP2]]
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] // CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK7: omp.arrayinit.done: // CHECK7: omp.arrayinit.done:
// CHECK7-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0 // CHECK7-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i32 0, i32 0
// CHECK7-NEXT: store ptr [[Z2]], ptr [[TMP5]], align 4 // CHECK7-NEXT: store ptr [[Z2]], ptr [[TMP3]], align 4
// CHECK7-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 // CHECK7-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4
// CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK7-NEXT: [[TMP10:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP8]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.9, ptr @.gomp_critical_user_.reduction.var) // CHECK7-NEXT: [[TMP6:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP5]], i32 1, i32 4, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @.omp.reduction.reduction_func.9, ptr @.gomp_critical_user_.reduction.var)
// CHECK7-NEXT: switch i32 [[TMP10]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ // CHECK7-NEXT: switch i32 [[TMP6]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK7-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] // CHECK7-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK7-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] // CHECK7-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK7-NEXT: ] // CHECK7-NEXT: ]
// CHECK7: .omp.reduction.case1: // CHECK7: .omp.reduction.case1:
// CHECK7-NEXT: [[TMP11:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99 // CHECK7-NEXT: [[TMP7:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99
// CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP11]] // CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP1]], [[TMP7]]
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE6:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK7: omp.arraycpy.body: // CHECK7: omp.arraycpy.body:
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST3:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT4:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK7-NEXT: [[TMP12:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK7-NEXT: [[TMP13:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP8]], [[TMP9]]
// CHECK7-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4 // CHECK7-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], align 4
// CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1 // CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT4]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST3]], i32 1
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK7-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP11]] // CHECK7-NEXT: [[OMP_ARRAYCPY_DONE5:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT4]], [[TMP7]]
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]] // CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_DONE6]], label [[OMP_ARRAYCPY_BODY]]
// CHECK7: omp.arraycpy.done6: // CHECK7: omp.arraycpy.done6:
// CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP8]], ptr @.gomp_critical_user_.reduction.var) // CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK7: .omp.reduction.case2: // CHECK7: .omp.reduction.case2:
// CHECK7-NEXT: [[TMP14:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99 // CHECK7-NEXT: [[TMP10:%.*]] = getelementptr i32, ptr [[TMP1]], i32 99
// CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP14]] // CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY7:%.*]] = icmp eq ptr [[TMP1]], [[TMP10]]
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]] // CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY7]], label [[OMP_ARRAYCPY_DONE14:%.*]], label [[OMP_ARRAYCPY_BODY8:%.*]]
// CHECK7: omp.arraycpy.body8: // CHECK7: omp.arraycpy.body8:
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ] // CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST9:%.*]] = phi ptr [ [[Z2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT12:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ] // CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[TMP1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT11:%.*]], [[OMP_ARRAYCPY_BODY8]] ]
// CHECK7-NEXT: [[TMP15:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4 // CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], align 4
// CHECK7-NEXT: [[TMP16:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP15]] monotonic, align 4 // CHECK7-NEXT: [[TMP12:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 [[TMP11]] monotonic, align 4
// CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1 // CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT11]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1 // CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT12]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST9]], i32 1
// CHECK7-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP14]] // CHECK7-NEXT: [[OMP_ARRAYCPY_DONE13:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT11]], [[TMP10]]
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]] // CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE13]], label [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_BODY8]]
// CHECK7: omp.arraycpy.done14: // CHECK7: omp.arraycpy.done14:
// CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP8]], ptr @.gomp_critical_user_.reduction.var) // CHECK7-NEXT: call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP5]], ptr @.gomp_critical_user_.reduction.var)
// CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] // CHECK7-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK7: .omp.reduction.default: // CHECK7: .omp.reduction.default:
// CHECK7-NEXT: ret void // CHECK7-NEXT: ret void
// //
// //
// CHECK7-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.9 // CHECK7-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func.9
// CHECK7-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] { // CHECK7-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR2]] {
// CHECK7-NEXT: entry: // CHECK7-NEXT: entry:
// CHECK7-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4 // CHECK7-NEXT: [[DOTADDR:%.*]] = alloca ptr, align 4
// CHECK7-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4 // CHECK7-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 4
// CHECK7-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4 // CHECK7-NEXT: store ptr [[TMP0]], ptr [[DOTADDR]], align 4
// CHECK7-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4 // CHECK7-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 4
// CHECK7-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4 // CHECK7-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 4
// CHECK7-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTADDR1]], align 4 // CHECK7-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 4
// CHECK7-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP4]], i32 0, i32 0 // CHECK7-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i32 0, i32 0
// CHECK7-NEXT: [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 4
// CHECK7-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0
// CHECK7-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4 // CHECK7-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 4
// CHECK7-NEXT: [[TMP9:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i32 0, i32 0 // CHECK7-NEXT: [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i32 99
// CHECK7-NEXT: [[TMP10:%.*]] = load ptr, ptr [[TMP9]], align 4 // CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// CHECK7-NEXT: [[TMP12:%.*]] = getelementptr i32, ptr [[TMP10]], i32 99
// CHECK7-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP10]], [[TMP12]]
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] // CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK7: omp.arraycpy.body: // CHECK7: omp.arraycpy.body:
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK7-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP10]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] // CHECK7-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK7-NEXT: [[TMP13:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK7-NEXT: [[TMP14:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4 // CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP13]], [[TMP14]] // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// CHECK7-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4 // CHECK7-NEXT: store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 // CHECK7-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 // CHECK7-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK7-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]] // CHECK7-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]] // CHECK7-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK7: omp.arraycpy.done2: // CHECK7: omp.arraycpy.done2:
// CHECK7-NEXT: ret void // CHECK7-NEXT: ret void
// //

llvm/include/llvm/Frontend/OpenMP/OMPIRBuilder.h

Show First 20 Lines • Show All 1,452 Lines • ▼ Show 20 Linespublic:
/// \see openmp/libomptarget/deviceRTLs/common/include/target.h /// \see openmp/libomptarget/deviceRTLs/common/include/target.h
/// ///
///{ ///{
/// Create a runtime call for kmpc_target_init /// Create a runtime call for kmpc_target_init
/// ///
/// \param Loc The insert and source location description. /// \param Loc The insert and source location description.
/// \param IsSPMD Flag to indicate if the kernel is an SPMD kernel or not. /// \param IsSPMD Flag to indicate if the kernel is an SPMD kernel or not.
/// \param RequiresFullRuntime Indicate if a full device runtime is necessary. InsertPointTy createTargetInit(const LocationDescription &Loc, bool IsSPMD);
InsertPointTy createTargetInit(const LocationDescription &Loc, bool IsSPMD,
bool RequiresFullRuntime);
/// Create a runtime call for kmpc_target_deinit /// Create a runtime call for kmpc_target_deinit
/// ///
/// \param Loc The insert and source location description. /// \param Loc The insert and source location description.
/// \param IsSPMD Flag to indicate if the kernel is an SPMD kernel or not. /// \param IsSPMD Flag to indicate if the kernel is an SPMD kernel or not.
/// \param RequiresFullRuntime Indicate if a full device runtime is necessary. void createTargetDeinit(const LocationDescription &Loc, bool IsSPMD);
void createTargetDeinit(const LocationDescription &Loc, bool IsSPMD,
bool RequiresFullRuntime);
///} ///}
private: private:
// Sets the function attributes expected for the outlined function // Sets the function attributes expected for the outlined function
void setOutlinedTargetRegionFunctionAttributes(Function *OutlinedFn, void setOutlinedTargetRegionFunctionAttributes(Function *OutlinedFn,
int32_t NumTeams, int32_t NumTeams,
int32_t NumThreads); int32_t NumThreads);
▲ Show 20 Lines • Show All 932 Lines • Show Last 20 Lines

llvm/include/llvm/Frontend/OpenMP/OMPKinds.def

Show First 20 Lines • Show All 441 Lines • ▼ Show 20 Lines__OMP_RTL(__tgt_target_data_update_nowait_mapper, false, Void, IdentPtr, Int64, Int32,
VoidPtrPtr, VoidPtrPtr, Int64Ptr, Int64Ptr, VoidPtrPtr, VoidPtrPtr) VoidPtrPtr, VoidPtrPtr, Int64Ptr, Int64Ptr, VoidPtrPtr, VoidPtrPtr)
__OMP_RTL(__tgt_mapper_num_components, false, Int64, VoidPtr) __OMP_RTL(__tgt_mapper_num_components, false, Int64, VoidPtr)
__OMP_RTL(__tgt_push_mapper_component, false, Void, VoidPtr, VoidPtr, VoidPtr, __OMP_RTL(__tgt_push_mapper_component, false, Void, VoidPtr, VoidPtr, VoidPtr,
Int64, Int64, VoidPtr) Int64, Int64, VoidPtr)
__OMP_RTL(__kmpc_task_allow_completion_event, false, VoidPtr, IdentPtr, __OMP_RTL(__kmpc_task_allow_completion_event, false, VoidPtr, IdentPtr,
/* Int */ Int32, /* kmp_task_t */ VoidPtr) /* Int */ Int32, /* kmp_task_t */ VoidPtr)
/// OpenMP Device runtime functions /// OpenMP Device runtime functions
__OMP_RTL(__kmpc_target_init, false, Int32, IdentPtr, Int8, Int1, Int1) __OMP_RTL(__kmpc_target_init, false, Int32, IdentPtr, Int8, Int1)
__OMP_RTL(__kmpc_target_deinit, false, Void, IdentPtr, Int8, Int1) __OMP_RTL(__kmpc_target_deinit, false, Void, IdentPtr, Int8)
__OMP_RTL(__kmpc_kernel_prepare_parallel, false, Void, VoidPtr) __OMP_RTL(__kmpc_kernel_prepare_parallel, false, Void, VoidPtr)
__OMP_RTL(__kmpc_parallel_51, false, Void, IdentPtr, Int32, Int32, Int32, Int32, __OMP_RTL(__kmpc_parallel_51, false, Void, IdentPtr, Int32, Int32, Int32, Int32,
VoidPtr, VoidPtr, VoidPtrPtr, SizeTy) VoidPtr, VoidPtr, VoidPtrPtr, SizeTy)
__OMP_RTL(__kmpc_kernel_parallel, false, Int1, VoidPtrPtr) __OMP_RTL(__kmpc_kernel_parallel, false, Int1, VoidPtrPtr)
__OMP_RTL(__kmpc_kernel_end_parallel, false, Void, ) __OMP_RTL(__kmpc_kernel_end_parallel, false, Void, )
__OMP_RTL(__kmpc_serialized_parallel, false, Void, IdentPtr, Int32) __OMP_RTL(__kmpc_serialized_parallel, false, Void, IdentPtr, Int32)
__OMP_RTL(__kmpc_end_serialized_parallel, false, Void, IdentPtr, Int32) __OMP_RTL(__kmpc_end_serialized_parallel, false, Void, IdentPtr, Int32)
__OMP_RTL(__kmpc_shuffle_int32, false, Int32, Int32, Int16, Int16) __OMP_RTL(__kmpc_shuffle_int32, false, Int32, Int32, Int16, Int16)
▲ Show 20 Lines • Show All 804 Lines • Show Last 20 Lines

llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp

Show First 20 Lines • Show All 3,848 Lines • ▼ Show 20 LinesCallInst *OpenMPIRBuilder::createCachedThreadPrivate(
Function *Fn = Function *Fn =
getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_threadprivate_cached); getOrCreateRuntimeFunctionPtr(OMPRTL___kmpc_threadprivate_cached);
return Builder.CreateCall(Fn, Args); return Builder.CreateCall(Fn, Args);
} }
OpenMPIRBuilder::InsertPointTy OpenMPIRBuilder::InsertPointTy
OpenMPIRBuilder::createTargetInit(const LocationDescription &Loc, bool IsSPMD, OpenMPIRBuilder::createTargetInit(const LocationDescription &Loc, bool IsSPMD) {
bool RequiresFullRuntime) {
if (!updateToLocation(Loc)) if (!updateToLocation(Loc))
return Loc.IP; return Loc.IP;
uint32_t SrcLocStrSize; uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize); Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Constant *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize); Constant *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
ConstantInt *IsSPMDVal = ConstantInt::getSigned( ConstantInt *IsSPMDVal = ConstantInt::getSigned(
IntegerType::getInt8Ty(Int8->getContext()), IntegerType::getInt8Ty(Int8->getContext()),
IsSPMD ? OMP_TGT_EXEC_MODE_SPMD : OMP_TGT_EXEC_MODE_GENERIC); IsSPMD ? OMP_TGT_EXEC_MODE_SPMD : OMP_TGT_EXEC_MODE_GENERIC);
ConstantInt *UseGenericStateMachine = ConstantInt *UseGenericStateMachine =
ConstantInt::getBool(Int32->getContext(), !IsSPMD); ConstantInt::getBool(Int32->getContext(), !IsSPMD);
ConstantInt *RequiresFullRuntimeVal =
ConstantInt::getBool(Int32->getContext(), RequiresFullRuntime);
Function *Fn = getOrCreateRuntimeFunctionPtr( Function *Fn = getOrCreateRuntimeFunctionPtr(
omp::RuntimeFunction::OMPRTL___kmpc_target_init); omp::RuntimeFunction::OMPRTL___kmpc_target_init);
CallInst *ThreadKind = Builder.CreateCall( CallInst *ThreadKind = Builder.CreateCall(
Fn, {Ident, IsSPMDVal, UseGenericStateMachine, RequiresFullRuntimeVal}); Fn, {Ident, IsSPMDVal, UseGenericStateMachine});
Value *ExecUserCode = Builder.CreateICmpEQ( Value *ExecUserCode = Builder.CreateICmpEQ(
ThreadKind, ConstantInt::get(ThreadKind->getType(), -1), ThreadKind, ConstantInt::get(ThreadKind->getType(), -1),
"exec_user_code"); "exec_user_code");
// ThreadKind = __kmpc_target_init(...) // ThreadKind = __kmpc_target_init(...)
// if (ThreadKind == -1) // if (ThreadKind == -1)
// user_code // user_code
Show All 17 LinesOpenMPIRBuilder::createTargetInit(const LocationDescription &Loc, bool IsSPMD) {
UI->eraseFromParent(); UI->eraseFromParent();
// Continue in the "user_code" block, see diagram above and in // Continue in the "user_code" block, see diagram above and in
// openmp/libomptarget/deviceRTLs/common/include/target.h . // openmp/libomptarget/deviceRTLs/common/include/target.h .
return InsertPointTy(UserCodeEntryBB, UserCodeEntryBB->getFirstInsertionPt()); return InsertPointTy(UserCodeEntryBB, UserCodeEntryBB->getFirstInsertionPt());
} }
void OpenMPIRBuilder::createTargetDeinit(const LocationDescription &Loc, void OpenMPIRBuilder::createTargetDeinit(const LocationDescription &Loc,
bool IsSPMD, bool IsSPMD) {
bool RequiresFullRuntime) {
if (!updateToLocation(Loc)) if (!updateToLocation(Loc))
return; return;
uint32_t SrcLocStrSize; uint32_t SrcLocStrSize;
Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize); Constant *SrcLocStr = getOrCreateSrcLocStr(Loc, SrcLocStrSize);
Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize); Value *Ident = getOrCreateIdent(SrcLocStr, SrcLocStrSize);
ConstantInt *IsSPMDVal = ConstantInt::getSigned( ConstantInt *IsSPMDVal = ConstantInt::getSigned(
IntegerType::getInt8Ty(Int8->getContext()), IntegerType::getInt8Ty(Int8->getContext()),
IsSPMD ? OMP_TGT_EXEC_MODE_SPMD : OMP_TGT_EXEC_MODE_GENERIC); IsSPMD ? OMP_TGT_EXEC_MODE_SPMD : OMP_TGT_EXEC_MODE_GENERIC);
ConstantInt *RequiresFullRuntimeVal =
ConstantInt::getBool(Int32->getContext(), RequiresFullRuntime);
Function *Fn = getOrCreateRuntimeFunctionPtr( Function *Fn = getOrCreateRuntimeFunctionPtr(
omp::RuntimeFunction::OMPRTL___kmpc_target_deinit); omp::RuntimeFunction::OMPRTL___kmpc_target_deinit);
Builder.CreateCall(Fn, {Ident, IsSPMDVal, RequiresFullRuntimeVal}); Builder.CreateCall(Fn, {Ident, IsSPMDVal});
} }
void OpenMPIRBuilder::setOutlinedTargetRegionFunctionAttributes( void OpenMPIRBuilder::setOutlinedTargetRegionFunctionAttributes(
Function *OutlinedFn, int32_t NumTeams, int32_t NumThreads) { Function *OutlinedFn, int32_t NumTeams, int32_t NumThreads) {
if (Config.isEmbedded()) { if (Config.isEmbedded()) {
OutlinedFn->setLinkage(GlobalValue::WeakODRLinkage); OutlinedFn->setLinkage(GlobalValue::WeakODRLinkage);
// TODO: Determine if DSO local can be set to true. // TODO: Determine if DSO local can be set to true.
OutlinedFn->setDSOLocal(false); OutlinedFn->setDSOLocal(false);
▲ Show 20 Lines • Show All 1,357 Lines • Show Last 20 Lines

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

Show First 20 Lines • Show All 1,857 Lines • ▼ Show 20 Linesprivate:
/// \p ReplVal if given. /// \p ReplVal if given.
bool deduplicateRuntimeCalls(Function &F, bool deduplicateRuntimeCalls(Function &F,
OMPInformationCache::RuntimeFunctionInfo &RFI, OMPInformationCache::RuntimeFunctionInfo &RFI,
Value *ReplVal = nullptr) { Value *ReplVal = nullptr) {
auto *UV = RFI.getUseVector(F); auto *UV = RFI.getUseVector(F);
if (!UV || UV->size() + (ReplVal != nullptr) < 2) if (!UV || UV->size() + (ReplVal != nullptr) < 2)
return false; return false;
LLVM_DEBUG( F.dump();
(
dbgs() << TAG << "Deduplicate " << UV->size() << " uses of " << RFI.Name dbgs() << TAG << "Deduplicate " << UV->size() << " uses of " << RFI.Name
<< (ReplVal ? " with an existing value\n" : "\n") << "\n"); << (ReplVal ? " with an existing value\n" : "\n") << "\n");
assert((!ReplVal || (isa<Argument>(ReplVal) && assert((!ReplVal || (isa<Argument>(ReplVal) &&
cast<Argument>(ReplVal)->getParent() == &F)) && cast<Argument>(ReplVal)->getParent() == &F)) &&
"Unexpected replacement value!"); "Unexpected replacement value!");
// TODO: Use dominance to find a good position instead. // TODO: Use dominance to find a good position instead.
▲ Show 20 Lines • Show All 1,410 Lines • ▼ Show 20 Lines Attributor::SimplifictionCallbackTy IsGenericModeSimplifyCB =
auto *Val = ConstantInt::getBool(IRP.getAnchorValue().getContext(), auto *Val = ConstantInt::getBool(IRP.getAnchorValue().getContext(),
!SPMDCompatibilityTracker.isAssumed()); !SPMDCompatibilityTracker.isAssumed());
return Val; return Val;
}; };
constexpr const int InitModeArgNo = 1; constexpr const int InitModeArgNo = 1;
constexpr const int DeinitModeArgNo = 1; constexpr const int DeinitModeArgNo = 1;
constexpr const int InitUseStateMachineArgNo = 2; constexpr const int InitUseStateMachineArgNo = 2;
constexpr const int InitRequiresFullRuntimeArgNo = 3;
constexpr const int DeinitRequiresFullRuntimeArgNo = 2;
A.registerSimplificationCallback( A.registerSimplificationCallback(
IRPosition::callsite_argument(*KernelInitCB, InitUseStateMachineArgNo), IRPosition::callsite_argument(*KernelInitCB, InitUseStateMachineArgNo),
StateMachineSimplifyCB); StateMachineSimplifyCB);
A.registerSimplificationCallback( A.registerSimplificationCallback(
IRPosition::callsite_argument(*KernelInitCB, InitModeArgNo), IRPosition::callsite_argument(*KernelInitCB, InitModeArgNo),
ModeSimplifyCB); ModeSimplifyCB);
A.registerSimplificationCallback( A.registerSimplificationCallback(
IRPosition::callsite_argument(*KernelDeinitCB, DeinitModeArgNo), IRPosition::callsite_argument(*KernelDeinitCB, DeinitModeArgNo),
ModeSimplifyCB); ModeSimplifyCB);
A.registerSimplificationCallback(
IRPosition::callsite_argument(*KernelInitCB,
InitRequiresFullRuntimeArgNo),
IsGenericModeSimplifyCB);
A.registerSimplificationCallback(
IRPosition::callsite_argument(*KernelDeinitCB,
DeinitRequiresFullRuntimeArgNo),
IsGenericModeSimplifyCB);
// Check if we know we are in SPMD-mode already. // Check if we know we are in SPMD-mode already.
ConstantInt *ModeArg = ConstantInt *ModeArg =
dyn_cast<ConstantInt>(KernelInitCB->getArgOperand(InitModeArgNo)); dyn_cast<ConstantInt>(KernelInitCB->getArgOperand(InitModeArgNo));
if (ModeArg && (ModeArg->getSExtValue() & OMP_TGT_EXEC_MODE_SPMD)) if (ModeArg && (ModeArg->getSExtValue() & OMP_TGT_EXEC_MODE_SPMD))
SPMDCompatibilityTracker.indicateOptimisticFixpoint(); SPMDCompatibilityTracker.indicateOptimisticFixpoint();
// This is a generic region but SPMDization is disabled so stop tracking. // This is a generic region but SPMDization is disabled so stop tracking.
else if (DisableOpenMPOptSPMDization) else if (DisableOpenMPOptSPMDization)
▲ Show 20 Lines • Show All 377 Lines • ▼ Show 20 Lines bool changeToSPMDMode(Attributor &A, ChangeStatus &Changed) {
ExecMode->setInitializer( ExecMode->setInitializer(
ConstantInt::get(ExecMode->getInitializer()->getType(), ConstantInt::get(ExecMode->getInitializer()->getType(),
ExecModeVal | OMP_TGT_EXEC_MODE_GENERIC_SPMD)); ExecModeVal | OMP_TGT_EXEC_MODE_GENERIC_SPMD));
// Next rewrite the init and deinit calls to indicate we use SPMD-mode now. // Next rewrite the init and deinit calls to indicate we use SPMD-mode now.
const int InitModeArgNo = 1; const int InitModeArgNo = 1;
const int DeinitModeArgNo = 1; const int DeinitModeArgNo = 1;
const int InitUseStateMachineArgNo = 2; const int InitUseStateMachineArgNo = 2;
const int InitRequiresFullRuntimeArgNo = 3;
const int DeinitRequiresFullRuntimeArgNo = 2;
auto &Ctx = getAnchorValue().getContext(); auto &Ctx = getAnchorValue().getContext();
A.changeUseAfterManifest( A.changeUseAfterManifest(
KernelInitCB->getArgOperandUse(InitModeArgNo), KernelInitCB->getArgOperandUse(InitModeArgNo),
*ConstantInt::getSigned(IntegerType::getInt8Ty(Ctx), *ConstantInt::getSigned(IntegerType::getInt8Ty(Ctx),
OMP_TGT_EXEC_MODE_SPMD)); OMP_TGT_EXEC_MODE_SPMD));
A.changeUseAfterManifest( A.changeUseAfterManifest(
KernelInitCB->getArgOperandUse(InitUseStateMachineArgNo), KernelInitCB->getArgOperandUse(InitUseStateMachineArgNo),
*ConstantInt::getBool(Ctx, false)); *ConstantInt::getBool(Ctx, false));
A.changeUseAfterManifest( A.changeUseAfterManifest(
KernelDeinitCB->getArgOperandUse(DeinitModeArgNo), KernelDeinitCB->getArgOperandUse(DeinitModeArgNo),
*ConstantInt::getSigned(IntegerType::getInt8Ty(Ctx), *ConstantInt::getSigned(IntegerType::getInt8Ty(Ctx),
OMP_TGT_EXEC_MODE_SPMD)); OMP_TGT_EXEC_MODE_SPMD));
A.changeUseAfterManifest(
KernelInitCB->getArgOperandUse(InitRequiresFullRuntimeArgNo),
*ConstantInt::getBool(Ctx, false));
A.changeUseAfterManifest(
KernelDeinitCB->getArgOperandUse(DeinitRequiresFullRuntimeArgNo),
*ConstantInt::getBool(Ctx, false));
++NumOpenMPTargetRegionKernelsSPMD; ++NumOpenMPTargetRegionKernelsSPMD;
auto Remark = [&](OptimizationRemark OR) { auto Remark = [&](OptimizationRemark OR) {
return OR << "Transformed generic-mode kernel to SPMD-mode."; return OR << "Transformed generic-mode kernel to SPMD-mode.";
}; };
A.emitRemark<OptimizationRemark>(KernelInitCB, "OMP120", Remark); A.emitRemark<OptimizationRemark>(KernelInitCB, "OMP120", Remark);
return true; return true;
▲ Show 20 Lines • Show All 1,476 Lines • Show Last 20 Lines

llvm/test/Transforms/OpenMP/always_inline_device.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-attributes ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-attributes
; RUN: opt < %s -S -passes=openmp-opt -openmp-opt-inline-device | FileCheck %s ; RUN: opt < %s -S -passes=openmp-opt -openmp-opt-inline-device | FileCheck %s
%struct.ident_t = type { i32, i32, i32, i32, i8* } %struct.ident_t = type { i32, i32, i32, i32, i8* }
@0 = private unnamed_addr constant [23 x i8] c";unknown;unknown;0;0;;\00", align 1 @0 = private unnamed_addr constant [23 x i8] c";unknown;unknown;0;0;;\00", align 1
@1 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* @0, i32 0, i32 0) }, align 8 @1 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* @0, i32 0, i32 0) }, align 8
@__omp_offloading_fd02_c0934fc2_foo_l4_exec_mode = weak constant i8 1 @__omp_offloading_fd02_c0934fc2_foo_l4_exec_mode = weak constant i8 1
@llvm.compiler.used = appending global [1 x i8*] [i8* @__omp_offloading_fd02_c0934fc2_foo_l4_exec_mode], section "llvm.metadata" @llvm.compiler.used = appending global [1 x i8*] [i8* @__omp_offloading_fd02_c0934fc2_foo_l4_exec_mode], section "llvm.metadata"
@G = external global i8 @G = external global i8
; Function Attrs: convergent norecurse nounwind ; Function Attrs: convergent norecurse nounwind
define weak void @__omp_offloading_fd02_c0934fc2_foo_l4() #0 { define weak void @__omp_offloading_fd02_c0934fc2_foo_l4() #0 {
; CHECK: Function Attrs: convergent norecurse nounwind ; CHECK: Function Attrs: convergent norecurse nounwind
; CHECK-LABEL: @__omp_offloading_fd02_c0934fc2_foo_l4( ; CHECK-LABEL: @__omp_offloading_fd02_c0934fc2_foo_l4(
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 false) ; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 2, i1 false)
; CHECK-NEXT: [[THREAD_ID_IN_BLOCK:%.*]] = call i32 @__kmpc_get_hardware_thread_id_in_block() ; CHECK-NEXT: [[THREAD_ID_IN_BLOCK:%.*]] = call i32 @__kmpc_get_hardware_thread_id_in_block()
; CHECK-NEXT: [[THREAD_IS_MAIN:%.*]] = icmp ne i32 [[THREAD_ID_IN_BLOCK]], 0 ; CHECK-NEXT: [[THREAD_IS_MAIN:%.*]] = icmp ne i32 [[THREAD_ID_IN_BLOCK]], 0
; CHECK-NEXT: br i1 [[THREAD_IS_MAIN]], label [[EXIT_THREADS:%.*]], label [[MAIN_THREAD_USER_CODE:%.*]] ; CHECK-NEXT: br i1 [[THREAD_IS_MAIN]], label [[EXIT_THREADS:%.*]], label [[MAIN_THREAD_USER_CODE:%.*]]
; CHECK: exit.threads: ; CHECK: exit.threads:
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; CHECK: main.thread.user_code: ; CHECK: main.thread.user_code:
; CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; 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-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; CHECK: user_code.entry: ; CHECK: user_code.entry:
; CHECK-NEXT: store i8 1, i8* @G, align 1 ; CHECK-NEXT: store i8 1, i8* @G, align 1
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 false) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; CHECK: worker.exit: ; CHECK: worker.exit:
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
entry: entry:
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
; Ensure we see a 0 here as the kernel doesn't have parallel regions and we want ; Ensure we see a 0 here as the kernel doesn't have parallel regions and we want
; generic execution. ; generic execution.
; TODO: This is not perfect. We should rather go for SPMD mode and tell the runtime ; TODO: This is not perfect. We should rather go for SPMD mode and tell the runtime
; to only spawn a single thread. Further, we then should not guard any code. ; to only spawn a single thread. Further, we then should not guard any code.
%isSPMD = call i8 @__kmpc_is_spmd_exec_mode() %isSPMD = call i8 @__kmpc_is_spmd_exec_mode()
store i8 %isSPMD, i8* @G store i8 %isSPMD, i8* @G
call void @bar() #2 call void @bar() #2
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
declare i8 @__kmpc_is_spmd_exec_mode() declare i8 @__kmpc_is_spmd_exec_mode()
declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1, i1) declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1)
declare void @__kmpc_target_deinit(%struct.ident_t*, i8, i1) declare void @__kmpc_target_deinit(%struct.ident_t*, i8)
; Function Attrs: convergent nounwind ; Function Attrs: convergent nounwind
define hidden void @bar() #1 { define hidden void @bar() #1 {
; CHECK: Function Attrs: alwaysinline convergent nounwind ; CHECK: Function Attrs: alwaysinline convergent nounwind
; CHECK-LABEL: @bar( ; CHECK-LABEL: @bar(
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
Show All 21 Lines

llvm/test/Transforms/OpenMP/barrier_removal.ll

Show First 20 Lines • Show All 243 Lines • ▼ Show 20 Lines
!10 = !{void ()* @neg_mem, !"kernel", i32 1} !10 = !{void ()* @neg_mem, !"kernel", i32 1}
!11 = !{void ()* @pos_multiple, !"kernel", i32 1} !11 = !{void ()* @pos_multiple, !"kernel", i32 1}
!12 = !{i32 7, !"openmp", i32 50} !12 = !{i32 7, !"openmp", i32 50}
!13 = !{i32 7, !"openmp-device", i32 50} !13 = !{i32 7, !"openmp-device", i32 50}
;. ;.
; CHECK: attributes #[[ATTR0:[0-9]+]] = { "llvm.assume"="ompx_aligned_barrier" } ; CHECK: attributes #[[ATTR0:[0-9]+]] = { "llvm.assume"="ompx_aligned_barrier" }
; CHECK: attributes #[[ATTR1:[0-9]+]] = { convergent nocallback nounwind } ; CHECK: attributes #[[ATTR1:[0-9]+]] = { convergent nocallback nounwind }
; CHECK: attributes #[[ATTR2:[0-9]+]] = { convergent nocallback nofree nounwind willreturn } ; CHECK: attributes #[[ATTR2:[0-9]+]] = { convergent nocallback nofree nounwind willreturn }
; CHECK: attributes #[[ATTR3:[0-9]+]] = { nocallback nofree nosync nounwind willreturn memory(inaccessiblemem: readwrite) }
;. ;.
; CHECK: [[META0:![0-9]+]] = !{i32 7, !"openmp", i32 50} ; CHECK: [[META0:![0-9]+]] = !{i32 7, !"openmp", i32 50}
; CHECK: [[META1:![0-9]+]] = !{i32 7, !"openmp-device", i32 50} ; CHECK: [[META1:![0-9]+]] = !{i32 7, !"openmp-device", i32 50}
; CHECK: [[META2:![0-9]+]] = !{void ()* @pos_empty_1, !"kernel", i32 1} ; CHECK: [[META2:![0-9]+]] = !{void ()* @pos_empty_1, !"kernel", i32 1}
; CHECK: [[META3:![0-9]+]] = !{void ()* @pos_empty_2, !"kernel", i32 1} ; CHECK: [[META3:![0-9]+]] = !{void ()* @pos_empty_2, !"kernel", i32 1}
; CHECK: [[META4:![0-9]+]] = !{void ()* @pos_empty_3, !"kernel", i32 1} ; CHECK: [[META4:![0-9]+]] = !{void ()* @pos_empty_3, !"kernel", i32 1}
; CHECK: [[META5:![0-9]+]] = !{void ()* @pos_empty_4, !"kernel", i32 1} ; CHECK: [[META5:![0-9]+]] = !{void ()* @pos_empty_4, !"kernel", i32 1}
; CHECK: [[META6:![0-9]+]] = !{void ()* @pos_empty_5, !"kernel", i32 1} ; CHECK: [[META6:![0-9]+]] = !{void ()* @pos_empty_5, !"kernel", i32 1}
; CHECK: [[META7:![0-9]+]] = !{void ()* @pos_empty_6, !"kernel", i32 1} ; CHECK: [[META7:![0-9]+]] = !{void ()* @pos_empty_6, !"kernel", i32 1}
; CHECK: [[META8:![0-9]+]] = !{void ()* @neg_empty_7, !"kernel", i32 1} ; CHECK: [[META8:![0-9]+]] = !{void ()* @neg_empty_7, !"kernel", i32 1}
; CHECK: [[META9:![0-9]+]] = !{void ()* @pos_constant_loads, !"kernel", i32 1} ; CHECK: [[META9:![0-9]+]] = !{void ()* @pos_constant_loads, !"kernel", i32 1}
; CHECK: [[META10:![0-9]+]] = !{void ()* @neg_loads, !"kernel", i32 1} ; CHECK: [[META10:![0-9]+]] = !{void ()* @neg_loads, !"kernel", i32 1}
; CHECK: [[META11:![0-9]+]] = !{void ()* @pos_priv_mem, !"kernel", i32 1} ; CHECK: [[META11:![0-9]+]] = !{void ()* @pos_priv_mem, !"kernel", i32 1}
; CHECK: [[META12:![0-9]+]] = !{void ()* @neg_mem, !"kernel", i32 1} ; CHECK: [[META12:![0-9]+]] = !{void ()* @neg_mem, !"kernel", i32 1}
; CHECK: [[META13:![0-9]+]] = !{void ()* @pos_multiple, !"kernel", i32 1} ; CHECK: [[META13:![0-9]+]] = !{void ()* @pos_multiple, !"kernel", i32 1}
;. ;.

llvm/test/Transforms/OpenMP/custom_state_machines.ll

Show First 20 Lines • Show All 135 Lines • ▼ Show 20 Lines
@3 = private unnamed_addr constant %struct.ident_t { i32 0, i32 322, i32 2, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* @0, i32 0, i32 0) }, align 8 @3 = private unnamed_addr constant %struct.ident_t { i32 0, i32 322, i32 2, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* @0, i32 0, i32 0) }, align 8
@llvm.compiler.used = appending global [8 x i8*] [i8* @__omp_offloading_14_a36502b_no_state_machine_needed_l14_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_l22_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_pure_l77_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92_exec_mode, i8* @__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112_exec_mode], section "llvm.metadata" @llvm.compiler.used = appending global [8 x i8*] [i8* @__omp_offloading_14_a36502b_no_state_machine_needed_l14_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_l22_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_pure_l77_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92_exec_mode, i8* @__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112_exec_mode], section "llvm.metadata"
define weak void @__omp_offloading_14_a36502b_no_state_machine_needed_l14() #0 { define weak void @__omp_offloading_14_a36502b_no_state_machine_needed_l14() #0 {
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 %1, i32* %.threadid_temp., align 4 store i32 %1, i32* %.threadid_temp., align 4
call void @__omp_outlined__(i32* %.threadid_temp., i32* %.zero.addr) #3 call void @__omp_outlined__(i32* %.threadid_temp., i32* %.zero.addr) #3
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
; Make it a weak definition so we will apply custom state machine rewriting but can't use the body in the reasoning. ; Make it a weak definition so we will apply custom state machine rewriting but can't use the body in the reasoning.
define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1, i1) { define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1) {
ret i32 0 ret i32 0
} }
define internal void @__omp_outlined__(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
entry: entry:
%.global_tid..addr = alloca i32*, align 8 %.global_tid..addr = alloca i32*, align 8
%.bound_tid..addr = alloca i32*, align 8 %.bound_tid..addr = alloca i32*, align 8
store i32* %.global_tid., i32** %.global_tid..addr, align 8 store i32* %.global_tid., i32** %.global_tid..addr, align 8
Show All 19 Linesomp_if.end: ; preds = %omp_if.then, %entry
call void @__kmpc_barrier(%struct.ident_t* @3, i32 %0) call void @__kmpc_barrier(%struct.ident_t* @3, i32 %0)
ret void ret void
} }
declare void @unknown_no_openmp() #2 declare void @unknown_no_openmp() #2
declare i32 @__kmpc_global_thread_num(%struct.ident_t*) #3 declare i32 @__kmpc_global_thread_num(%struct.ident_t*) #3
declare void @__kmpc_target_deinit(%struct.ident_t*, i8, i1) declare void @__kmpc_target_deinit(%struct.ident_t*, i8)
define weak void @__omp_offloading_14_a36502b_simple_state_machine_l22() #0 { define weak void @__omp_offloading_14_a36502b_simple_state_machine_l22() #0 {
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 %1, i32* %.threadid_temp., align 4 store i32 %1, i32* %.threadid_temp., align 4
call void @__omp_outlined__1(i32* %.threadid_temp., i32* %.zero.addr) #3 call void @__omp_outlined__1(i32* %.threadid_temp., i32* %.zero.addr) #3
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
define internal void @__omp_outlined__1(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__1(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
entry: entry:
▲ Show 20 Lines • Show All 70 Lines • ▼ Show 20 Linesentry:
ret void ret void
} }
define weak void @__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39() #0 { define weak void @__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39() #0 {
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 %1, i32* %.threadid_temp., align 4 store i32 %1, i32* %.threadid_temp., align 4
call void @__omp_outlined__4(i32* %.threadid_temp., i32* %.zero.addr) #3 call void @__omp_outlined__4(i32* %.threadid_temp., i32* %.zero.addr) #3
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
define internal void @__omp_outlined__4(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__4(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
entry: entry:
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Linesentry:
ret void ret void
} }
define weak void @__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55() #0 { define weak void @__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55() #0 {
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 %1, i32* %.threadid_temp., align 4 store i32 %1, i32* %.threadid_temp., align 4
call void @__omp_outlined__6(i32* %.threadid_temp., i32* %.zero.addr) #3 call void @__omp_outlined__6(i32* %.threadid_temp., i32* %.zero.addr) #3
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
define internal void @__omp_outlined__6(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__6(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
entry: entry:
▲ Show 20 Lines • Show All 63 Lines • ▼ Show 20 Linesentry:
ret void ret void
} }
define weak void @__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66() #0 { define weak void @__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66() #0 {
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 %1, i32* %.threadid_temp., align 4 store i32 %1, i32* %.threadid_temp., align 4
call void @__omp_outlined__9(i32* %.threadid_temp., i32* %.zero.addr) #3 call void @__omp_outlined__9(i32* %.threadid_temp., i32* %.zero.addr) #3
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
define internal void @__omp_outlined__9(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__9(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
entry: entry:
▲ Show 20 Lines • Show All 61 Lines • ▼ Show 20 Linesentry:
ret void ret void
} }
define weak void @__omp_offloading_14_a36502b_simple_state_machine_pure_l77() #0 { define weak void @__omp_offloading_14_a36502b_simple_state_machine_pure_l77() #0 {
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 %1, i32* %.threadid_temp., align 4 store i32 %1, i32* %.threadid_temp., align 4
call void @__omp_outlined__12(i32* %.threadid_temp., i32* %.zero.addr) #3 call void @__omp_outlined__12(i32* %.threadid_temp., i32* %.zero.addr) #3
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
define internal void @__omp_outlined__12(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__12(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
entry: entry:
▲ Show 20 Lines • Show All 64 Lines • ▼ Show 20 Linesentry:
ret void ret void
} }
define weak void @__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92() #0 { define weak void @__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92() #0 {
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 %1, i32* %.threadid_temp., align 4 store i32 %1, i32* %.threadid_temp., align 4
call void @__omp_outlined__15(i32* %.threadid_temp., i32* %.zero.addr) #3 call void @__omp_outlined__15(i32* %.threadid_temp., i32* %.zero.addr) #3
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
define internal void @__omp_outlined__15(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__15(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
entry: entry:
Show All 30 Lines
declare i32 @omp_get_thread_num(...) #4 declare i32 @omp_get_thread_num(...) #4
define weak void @__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112() #0 { define weak void @__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112() #0 {
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 %1, i32* %.threadid_temp., align 4 store i32 %1, i32* %.threadid_temp., align 4
call void @__omp_outlined__16(i32* %.threadid_temp., i32* %.zero.addr) #3 call void @__omp_outlined__16(i32* %.threadid_temp., i32* %.zero.addr) #3
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
define internal void @__omp_outlined__16(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__16(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
entry: entry:
▲ Show 20 Lines • Show All 132 Lines • ▼ Show 20 Lines
!17 = !{i32 7, !"openmp", i32 50} !17 = !{i32 7, !"openmp", i32 50}
!18 = !{i32 7, !"openmp-device", i32 50} !18 = !{i32 7, !"openmp-device", i32 50}
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_needed_l14 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_needed_l14
; AMDGPU-SAME: () #[[ATTR0:[0-9]+]] { ; AMDGPU-SAME: () #[[ATTR0:[0-9]+]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 1, i1 false, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 1, i1 false)
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3:[0-9]+]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3:[0-9]+]]
; AMDGPU-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: worker.exit: ; AMDGPU: worker.exit:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; ;
; ;
; AMDGPU-LABEL: define {{[^@]+}}@__kmpc_target_init
; AMDGPU-SAME: (%struct.ident_t* [[TMP0:%.*]], i8 [[TMP1:%.*]], i1 [[TMP2:%.*]]) {
; AMDGPU-NEXT: ret i32 0
;
;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__ ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 ; AMDGPU-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
; AMDGPU-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 ; AMDGPU-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
; AMDGPU-NEXT: call void @no_parallel_region_in_here.internalized() #[[ATTR8:[0-9]+]] ; AMDGPU-NEXT: call void @no_parallel_region_in_here.internalized() #[[ATTR8:[0-9]+]]
; AMDGPU-NEXT: call void @unknown_no_openmp() #[[ATTR9:[0-9]+]] ; AMDGPU-NEXT: call void @unknown_no_openmp() #[[ATTR9:[0-9]+]]
Show All 36 Lines
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_l22 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_l22
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5) ; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5)
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; AMDGPU: is_worker_check: ; AMDGPU: is_worker_check:
; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 30 Lines
; AMDGPU-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]] ; AMDGPU-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]]
; AMDGPU: thread.user_code.check: ; AMDGPU: thread.user_code.check:
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; AMDGPU-NEXT: call void @__omp_outlined__1(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-NEXT: call void @__omp_outlined__1(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: worker.exit: ; AMDGPU: worker.exit:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__1 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__1
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 62 Lines • ▼ Show 20 Lines
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5) ; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5)
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; AMDGPU: is_worker_check: ; AMDGPU: is_worker_check:
; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 36 Lines
; AMDGPU-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]] ; AMDGPU-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]]
; AMDGPU: thread.user_code.check: ; AMDGPU: thread.user_code.check:
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; AMDGPU-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: worker.exit: ; AMDGPU: worker.exit:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__4 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__4
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 82 Lines • ▼ Show 20 Lines
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5) ; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5)
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; AMDGPU: is_worker_check: ; AMDGPU: is_worker_check:
; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 32 Lines
; AMDGPU-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]] ; AMDGPU-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]]
; AMDGPU: thread.user_code.check: ; AMDGPU: thread.user_code.check:
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; AMDGPU-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: worker.exit: ; AMDGPU: worker.exit:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__6 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__6
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 61 Lines • ▼ Show 20 Lines
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5) ; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5)
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; AMDGPU: is_worker_check: ; AMDGPU: is_worker_check:
; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 30 Lines
; AMDGPU-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]] ; AMDGPU-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]]
; AMDGPU: thread.user_code.check: ; AMDGPU: thread.user_code.check:
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; AMDGPU-NEXT: call void @__omp_outlined__9(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-NEXT: call void @__omp_outlined__9(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: worker.exit: ; AMDGPU: worker.exit:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__9 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__9
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 61 Lines • ▼ Show 20 Lines
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_pure_l77 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_pure_l77
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5) ; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5)
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; AMDGPU: is_worker_check: ; AMDGPU: is_worker_check:
; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 30 Lines
; AMDGPU-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]] ; AMDGPU-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]]
; AMDGPU: thread.user_code.check: ; AMDGPU: thread.user_code.check:
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; AMDGPU-NEXT: call void @__omp_outlined__12(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-NEXT: call void @__omp_outlined__12(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: worker.exit: ; AMDGPU: worker.exit:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__12 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__12
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 61 Lines • ▼ Show 20 Lines
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5) ; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5)
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; AMDGPU: is_worker_check: ; AMDGPU: is_worker_check:
; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 25 Lines
; AMDGPU-NEXT: call void @__kmpc_barrier_simple_generic(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]]) ; AMDGPU-NEXT: call void @__kmpc_barrier_simple_generic(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
; AMDGPU-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]] ; AMDGPU-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]]
; AMDGPU: thread.user_code.check: ; AMDGPU: thread.user_code.check:
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__omp_outlined__15(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-NEXT: call void @__omp_outlined__15(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: worker.exit: ; AMDGPU: worker.exit:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__15 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__15
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 49 Lines • ▼ Show 20 Lines
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5) ; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5)
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; AMDGPU: is_worker_check: ; AMDGPU: is_worker_check:
; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 19 Lines
; AMDGPU-NEXT: call void @__kmpc_barrier_simple_generic(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]]) ; AMDGPU-NEXT: call void @__kmpc_barrier_simple_generic(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
; AMDGPU-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]] ; AMDGPU-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]]
; AMDGPU: thread.user_code.check: ; AMDGPU: thread.user_code.check:
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__omp_outlined__16(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-NEXT: call void @__omp_outlined__16(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: worker.exit: ; AMDGPU: worker.exit:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__16 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__16
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 103 Lines • ▼ Show 20 Lines
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_needed_l14 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_needed_l14
; NVPTX-SAME: () #[[ATTR0:[0-9]+]] { ; NVPTX-SAME: () #[[ATTR0:[0-9]+]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 1, i1 false, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 1, i1 false)
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3:[0-9]+]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3:[0-9]+]]
; NVPTX-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: worker.exit: ; NVPTX: worker.exit:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; ;
; ;
; NVPTX-LABEL: define {{[^@]+}}@__kmpc_target_init
; NVPTX-SAME: (%struct.ident_t* [[TMP0:%.*]], i8 [[TMP1:%.*]], i1 [[TMP2:%.*]]) {
; NVPTX-NEXT: ret i32 0
;
;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__ ; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__
; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 ; NVPTX-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
; NVPTX-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 ; NVPTX-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
; NVPTX-NEXT: call void @no_parallel_region_in_here.internalized() #[[ATTR8:[0-9]+]] ; NVPTX-NEXT: call void @no_parallel_region_in_here.internalized() #[[ATTR8:[0-9]+]]
; NVPTX-NEXT: call void @unknown_no_openmp() #[[ATTR9:[0-9]+]] ; NVPTX-NEXT: call void @unknown_no_openmp() #[[ATTR9:[0-9]+]]
Show All 36 Lines
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_l22 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_l22
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8 ; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; NVPTX: is_worker_check: ; NVPTX: is_worker_check:
; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 29 Lines
; NVPTX-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]] ; NVPTX-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]]
; NVPTX: thread.user_code.check: ; NVPTX: thread.user_code.check:
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; NVPTX-NEXT: call void @__omp_outlined__1(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-NEXT: call void @__omp_outlined__1(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: worker.exit: ; NVPTX: worker.exit:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__1 ; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__1
; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 62 Lines • ▼ Show 20 Lines
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8 ; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; NVPTX: is_worker_check: ; NVPTX: is_worker_check:
; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 35 Lines
; NVPTX-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]] ; NVPTX-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]]
; NVPTX: thread.user_code.check: ; NVPTX: thread.user_code.check:
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; NVPTX-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: worker.exit: ; NVPTX: worker.exit:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__4 ; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__4
; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 82 Lines • ▼ Show 20 Lines
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8 ; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; NVPTX: is_worker_check: ; NVPTX: is_worker_check:
; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 31 Lines
; NVPTX-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]] ; NVPTX-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]]
; NVPTX: thread.user_code.check: ; NVPTX: thread.user_code.check:
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; NVPTX-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: worker.exit: ; NVPTX: worker.exit:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__6 ; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__6
; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 61 Lines • ▼ Show 20 Lines
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8 ; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; NVPTX: is_worker_check: ; NVPTX: is_worker_check:
; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 29 Lines
; NVPTX-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]] ; NVPTX-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]]
; NVPTX: thread.user_code.check: ; NVPTX: thread.user_code.check:
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; NVPTX-NEXT: call void @__omp_outlined__9(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-NEXT: call void @__omp_outlined__9(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: worker.exit: ; NVPTX: worker.exit:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__9 ; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__9
; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 61 Lines • ▼ Show 20 Lines
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_pure_l77 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_pure_l77
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8 ; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; NVPTX: is_worker_check: ; NVPTX: is_worker_check:
; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 29 Lines
; NVPTX-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]] ; NVPTX-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]]
; NVPTX: thread.user_code.check: ; NVPTX: thread.user_code.check:
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; NVPTX-NEXT: call void @__omp_outlined__12(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-NEXT: call void @__omp_outlined__12(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: worker.exit: ; NVPTX: worker.exit:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__12 ; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__12
; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 61 Lines • ▼ Show 20 Lines
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8 ; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; NVPTX: is_worker_check: ; NVPTX: is_worker_check:
; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 24 Lines
; NVPTX-NEXT: call void @__kmpc_barrier_simple_generic(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]]) ; NVPTX-NEXT: call void @__kmpc_barrier_simple_generic(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
; NVPTX-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]] ; NVPTX-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]]
; NVPTX: thread.user_code.check: ; NVPTX: thread.user_code.check:
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__omp_outlined__15(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-NEXT: call void @__omp_outlined__15(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: worker.exit: ; NVPTX: worker.exit:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__15 ; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__15
; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 49 Lines • ▼ Show 20 Lines
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8 ; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; NVPTX: is_worker_check: ; NVPTX: is_worker_check:
; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 18 Lines
; NVPTX-NEXT: call void @__kmpc_barrier_simple_generic(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]]) ; NVPTX-NEXT: call void @__kmpc_barrier_simple_generic(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
; NVPTX-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]] ; NVPTX-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]]
; NVPTX: thread.user_code.check: ; NVPTX: thread.user_code.check:
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__omp_outlined__16(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-NEXT: call void @__omp_outlined__16(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: worker.exit: ; NVPTX: worker.exit:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__16 ; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__16
; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 103 Lines • ▼ Show 20 Lines
; ;
; ;
; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_needed_l14 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_needed_l14
; AMDGPU-DISABLED-SAME: () #[[ATTR0:[0-9]+]] { ; AMDGPU-DISABLED-SAME: () #[[ATTR0:[0-9]+]] {
; AMDGPU-DISABLED-NEXT: entry: ; AMDGPU-DISABLED-NEXT: entry:
; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) ; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 1, i1 true)
; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU-DISABLED: user_code.entry: ; AMDGPU-DISABLED: user_code.entry:
; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3:[0-9]+]] ; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3:[0-9]+]]
; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; AMDGPU-DISABLED: worker.exit: ; AMDGPU-DISABLED: worker.exit:
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; ;
; ;
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__kmpc_target_init
; AMDGPU-DISABLED-SAME: (%struct.ident_t* [[TMP0:%.*]], i8 [[TMP1:%.*]], i1 [[TMP2:%.*]]) {
; AMDGPU-DISABLED-NEXT: ret i32 0
;
;
; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__ ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__
; AMDGPU-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
; AMDGPU-DISABLED-NEXT: entry: ; AMDGPU-DISABLED-NEXT: entry:
; AMDGPU-DISABLED-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 ; AMDGPU-DISABLED-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
; AMDGPU-DISABLED-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 ; AMDGPU-DISABLED-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
; AMDGPU-DISABLED-NEXT: call void @no_parallel_region_in_here.internalized() #[[ATTR8:[0-9]+]] ; AMDGPU-DISABLED-NEXT: call void @no_parallel_region_in_here.internalized() #[[ATTR8:[0-9]+]]
; AMDGPU-DISABLED-NEXT: call void @unknown_no_openmp() #[[ATTR9:[0-9]+]] ; AMDGPU-DISABLED-NEXT: call void @unknown_no_openmp() #[[ATTR9:[0-9]+]]
Show All 35 Lines
; ;
; ;
; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_l22 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_l22
; AMDGPU-DISABLED-SAME: () #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: () #[[ATTR0]] {
; AMDGPU-DISABLED-NEXT: entry: ; AMDGPU-DISABLED-NEXT: entry:
; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU-DISABLED: user_code.entry: ; AMDGPU-DISABLED: user_code.entry:
; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__1(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__1(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; AMDGPU-DISABLED: worker.exit: ; AMDGPU-DISABLED: worker.exit:
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; ;
; ;
; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__1 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__1
; AMDGPU-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 61 Lines • ▼ Show 20 Lines
; ;
; ;
; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39
; AMDGPU-DISABLED-SAME: () #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: () #[[ATTR0]] {
; AMDGPU-DISABLED-NEXT: entry: ; AMDGPU-DISABLED-NEXT: entry:
; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU-DISABLED: user_code.entry: ; AMDGPU-DISABLED: user_code.entry:
; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; AMDGPU-DISABLED: worker.exit: ; AMDGPU-DISABLED: worker.exit:
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; ;
; ;
; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__4 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__4
; AMDGPU-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 81 Lines • ▼ Show 20 Lines
; ;
; ;
; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55
; AMDGPU-DISABLED-SAME: () #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: () #[[ATTR0]] {
; AMDGPU-DISABLED-NEXT: entry: ; AMDGPU-DISABLED-NEXT: entry:
; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU-DISABLED: user_code.entry: ; AMDGPU-DISABLED: user_code.entry:
; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; AMDGPU-DISABLED: worker.exit: ; AMDGPU-DISABLED: worker.exit:
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; ;
; ;
; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__6 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__6
; AMDGPU-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
; ;
; ;
; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66
; AMDGPU-DISABLED-SAME: () #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: () #[[ATTR0]] {
; AMDGPU-DISABLED-NEXT: entry: ; AMDGPU-DISABLED-NEXT: entry:
; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU-DISABLED: user_code.entry: ; AMDGPU-DISABLED: user_code.entry:
; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__9(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__9(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; AMDGPU-DISABLED: worker.exit: ; AMDGPU-DISABLED: worker.exit:
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; ;
; ;
; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__9 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__9
; AMDGPU-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
; ;
; ;
; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_pure_l77 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_pure_l77
; AMDGPU-DISABLED-SAME: () #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: () #[[ATTR0]] {
; AMDGPU-DISABLED-NEXT: entry: ; AMDGPU-DISABLED-NEXT: entry:
; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU-DISABLED: user_code.entry: ; AMDGPU-DISABLED: user_code.entry:
; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__12(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__12(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; AMDGPU-DISABLED: worker.exit: ; AMDGPU-DISABLED: worker.exit:
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; ;
; ;
; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__12 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__12
; AMDGPU-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
; ;
; ;
; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92
; AMDGPU-DISABLED-SAME: () #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: () #[[ATTR0]] {
; AMDGPU-DISABLED-NEXT: entry: ; AMDGPU-DISABLED-NEXT: entry:
; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU-DISABLED: user_code.entry: ; AMDGPU-DISABLED: user_code.entry:
; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__15(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__15(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; AMDGPU-DISABLED: worker.exit: ; AMDGPU-DISABLED: worker.exit:
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; ;
; ;
; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__15 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__15
; AMDGPU-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
; ;
; ;
; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112
; AMDGPU-DISABLED-SAME: () #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: () #[[ATTR0]] {
; AMDGPU-DISABLED-NEXT: entry: ; AMDGPU-DISABLED-NEXT: entry:
; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU-DISABLED: user_code.entry: ; AMDGPU-DISABLED: user_code.entry:
; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__16(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__16(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; AMDGPU-DISABLED: worker.exit: ; AMDGPU-DISABLED: worker.exit:
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; ;
; ;
; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__16 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__16
; AMDGPU-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 103 Lines • ▼ Show 20 Lines
; ;
; ;
; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_needed_l14 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_needed_l14
; NVPTX-DISABLED-SAME: () #[[ATTR0:[0-9]+]] { ; NVPTX-DISABLED-SAME: () #[[ATTR0:[0-9]+]] {
; NVPTX-DISABLED-NEXT: entry: ; NVPTX-DISABLED-NEXT: entry:
; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) ; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 1, i1 true)
; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX-DISABLED: user_code.entry: ; NVPTX-DISABLED: user_code.entry:
; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3:[0-9]+]] ; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3:[0-9]+]]
; NVPTX-DISABLED-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-DISABLED-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; NVPTX-DISABLED: worker.exit: ; NVPTX-DISABLED: worker.exit:
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; ;
; ;
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__kmpc_target_init
; NVPTX-DISABLED-SAME: (%struct.ident_t* [[TMP0:%.*]], i8 [[TMP1:%.*]], i1 [[TMP2:%.*]]) {
; NVPTX-DISABLED-NEXT: ret i32 0
;
;
; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__ ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__
; NVPTX-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
; NVPTX-DISABLED-NEXT: entry: ; NVPTX-DISABLED-NEXT: entry:
; NVPTX-DISABLED-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 ; NVPTX-DISABLED-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8
; NVPTX-DISABLED-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 ; NVPTX-DISABLED-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8
; NVPTX-DISABLED-NEXT: call void @no_parallel_region_in_here.internalized() #[[ATTR8:[0-9]+]] ; NVPTX-DISABLED-NEXT: call void @no_parallel_region_in_here.internalized() #[[ATTR8:[0-9]+]]
; NVPTX-DISABLED-NEXT: call void @unknown_no_openmp() #[[ATTR9:[0-9]+]] ; NVPTX-DISABLED-NEXT: call void @unknown_no_openmp() #[[ATTR9:[0-9]+]]
Show All 35 Lines
; ;
; ;
; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_l22 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_l22
; NVPTX-DISABLED-SAME: () #[[ATTR0]] { ; NVPTX-DISABLED-SAME: () #[[ATTR0]] {
; NVPTX-DISABLED-NEXT: entry: ; NVPTX-DISABLED-NEXT: entry:
; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX-DISABLED: user_code.entry: ; NVPTX-DISABLED: user_code.entry:
; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; NVPTX-DISABLED-NEXT: call void @__omp_outlined__1(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-DISABLED-NEXT: call void @__omp_outlined__1(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; NVPTX-DISABLED: worker.exit: ; NVPTX-DISABLED: worker.exit:
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; ;
; ;
; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__1 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__1
; NVPTX-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 61 Lines • ▼ Show 20 Lines
; ;
; ;
; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39
; NVPTX-DISABLED-SAME: () #[[ATTR0]] { ; NVPTX-DISABLED-SAME: () #[[ATTR0]] {
; NVPTX-DISABLED-NEXT: entry: ; NVPTX-DISABLED-NEXT: entry:
; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX-DISABLED: user_code.entry: ; NVPTX-DISABLED: user_code.entry:
; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; NVPTX-DISABLED-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-DISABLED-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; NVPTX-DISABLED: worker.exit: ; NVPTX-DISABLED: worker.exit:
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; ;
; ;
; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__4 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__4
; NVPTX-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 81 Lines • ▼ Show 20 Lines
; ;
; ;
; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55
; NVPTX-DISABLED-SAME: () #[[ATTR0]] { ; NVPTX-DISABLED-SAME: () #[[ATTR0]] {
; NVPTX-DISABLED-NEXT: entry: ; NVPTX-DISABLED-NEXT: entry:
; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX-DISABLED: user_code.entry: ; NVPTX-DISABLED: user_code.entry:
; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; NVPTX-DISABLED-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-DISABLED-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; NVPTX-DISABLED: worker.exit: ; NVPTX-DISABLED: worker.exit:
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; ;
; ;
; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__6 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__6
; NVPTX-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
; ;
; ;
; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66
; NVPTX-DISABLED-SAME: () #[[ATTR0]] { ; NVPTX-DISABLED-SAME: () #[[ATTR0]] {
; NVPTX-DISABLED-NEXT: entry: ; NVPTX-DISABLED-NEXT: entry:
; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX-DISABLED: user_code.entry: ; NVPTX-DISABLED: user_code.entry:
; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; NVPTX-DISABLED-NEXT: call void @__omp_outlined__9(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-DISABLED-NEXT: call void @__omp_outlined__9(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; NVPTX-DISABLED: worker.exit: ; NVPTX-DISABLED: worker.exit:
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; ;
; ;
; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__9 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__9
; NVPTX-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
; ;
; ;
; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_pure_l77 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_pure_l77
; NVPTX-DISABLED-SAME: () #[[ATTR0]] { ; NVPTX-DISABLED-SAME: () #[[ATTR0]] {
; NVPTX-DISABLED-NEXT: entry: ; NVPTX-DISABLED-NEXT: entry:
; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX-DISABLED: user_code.entry: ; NVPTX-DISABLED: user_code.entry:
; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; NVPTX-DISABLED-NEXT: call void @__omp_outlined__12(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-DISABLED-NEXT: call void @__omp_outlined__12(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; NVPTX-DISABLED: worker.exit: ; NVPTX-DISABLED: worker.exit:
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; ;
; ;
; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__12 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__12
; NVPTX-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
; ;
; ;
; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92
; NVPTX-DISABLED-SAME: () #[[ATTR0]] { ; NVPTX-DISABLED-SAME: () #[[ATTR0]] {
; NVPTX-DISABLED-NEXT: entry: ; NVPTX-DISABLED-NEXT: entry:
; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX-DISABLED: user_code.entry: ; NVPTX-DISABLED: user_code.entry:
; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-DISABLED-NEXT: call void @__omp_outlined__15(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-DISABLED-NEXT: call void @__omp_outlined__15(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; NVPTX-DISABLED: worker.exit: ; NVPTX-DISABLED: worker.exit:
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; ;
; ;
; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__15 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__15
; NVPTX-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
; ;
; ;
; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112
; NVPTX-DISABLED-SAME: () #[[ATTR0]] { ; NVPTX-DISABLED-SAME: () #[[ATTR0]] {
; NVPTX-DISABLED-NEXT: entry: ; NVPTX-DISABLED-NEXT: entry:
; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX-DISABLED: user_code.entry: ; NVPTX-DISABLED: user_code.entry:
; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-DISABLED-NEXT: call void @__omp_outlined__16(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-DISABLED-NEXT: call void @__omp_outlined__16(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; NVPTX-DISABLED: worker.exit: ; NVPTX-DISABLED: worker.exit:
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; ;
; ;
; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind ; NVPTX-DISABLED: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__16 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_outlined__16
; NVPTX-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-DISABLED-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 104 Lines • Show Last 20 Lines

llvm/test/Transforms/OpenMP/custom_state_machines_pre_lto.ll

Show First 20 Lines • Show All 135 Lines • ▼ Show 20 Lines
@3 = private unnamed_addr constant %struct.ident_t { i32 0, i32 322, i32 2, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* @0, i32 0, i32 0) }, align 8 @3 = private unnamed_addr constant %struct.ident_t { i32 0, i32 322, i32 2, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* @0, i32 0, i32 0) }, align 8
@llvm.compiler.used = appending global [8 x i8*] [i8* @__omp_offloading_14_a36502b_no_state_machine_needed_l14_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_l22_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_pure_l77_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92_exec_mode, i8* @__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112_exec_mode], section "llvm.metadata" @llvm.compiler.used = appending global [8 x i8*] [i8* @__omp_offloading_14_a36502b_no_state_machine_needed_l14_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_l22_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_pure_l77_exec_mode, i8* @__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92_exec_mode, i8* @__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112_exec_mode], section "llvm.metadata"
define weak void @__omp_offloading_14_a36502b_no_state_machine_needed_l14() #0 { define weak void @__omp_offloading_14_a36502b_no_state_machine_needed_l14() #0 {
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 %1, i32* %.threadid_temp., align 4 store i32 %1, i32* %.threadid_temp., align 4
call void @__omp_outlined__(i32* %.threadid_temp., i32* %.zero.addr) #3 call void @__omp_outlined__(i32* %.threadid_temp., i32* %.zero.addr) #3
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
; Make it a declaration so we will *not* apply custom state machine rewriting and wait for LTO. ; Make it a declaration so we will *not* apply custom state machine rewriting and wait for LTO.
declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1, i1); declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1);
define internal void @__omp_outlined__(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
entry: entry:
%.global_tid..addr = alloca i32*, align 8 %.global_tid..addr = alloca i32*, align 8
%.bound_tid..addr = alloca i32*, align 8 %.bound_tid..addr = alloca i32*, align 8
store i32* %.global_tid., i32** %.global_tid..addr, align 8 store i32* %.global_tid., i32** %.global_tid..addr, align 8
store i32* %.bound_tid., i32** %.bound_tid..addr, align 8 store i32* %.bound_tid., i32** %.bound_tid..addr, align 8
call void @no_parallel_region_in_here() #7 call void @no_parallel_region_in_here() #7
Show All 17 Linesomp_if.end: ; preds = %omp_if.then, %entry
call void @__kmpc_barrier(%struct.ident_t* @3, i32 %0) call void @__kmpc_barrier(%struct.ident_t* @3, i32 %0)
ret void ret void
} }
declare void @unknown_no_openmp() #2 declare void @unknown_no_openmp() #2
declare i32 @__kmpc_global_thread_num(%struct.ident_t*) #3 declare i32 @__kmpc_global_thread_num(%struct.ident_t*) #3
declare void @__kmpc_target_deinit(%struct.ident_t*, i8, i1) declare void @__kmpc_target_deinit(%struct.ident_t*, i8)
define weak void @__omp_offloading_14_a36502b_simple_state_machine_l22() #0 { define weak void @__omp_offloading_14_a36502b_simple_state_machine_l22() #0 {
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 %1, i32* %.threadid_temp., align 4 store i32 %1, i32* %.threadid_temp., align 4
call void @__omp_outlined__1(i32* %.threadid_temp., i32* %.zero.addr) #3 call void @__omp_outlined__1(i32* %.threadid_temp., i32* %.zero.addr) #3
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
define internal void @__omp_outlined__1(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__1(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
entry: entry:
▲ Show 20 Lines • Show All 70 Lines • ▼ Show 20 Linesentry:
ret void ret void
} }
define weak void @__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39() #0 { define weak void @__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39() #0 {
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 %1, i32* %.threadid_temp., align 4 store i32 %1, i32* %.threadid_temp., align 4
call void @__omp_outlined__4(i32* %.threadid_temp., i32* %.zero.addr) #3 call void @__omp_outlined__4(i32* %.threadid_temp., i32* %.zero.addr) #3
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
define internal void @__omp_outlined__4(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__4(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
entry: entry:
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Linesentry:
ret void ret void
} }
define weak void @__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55() #0 { define weak void @__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55() #0 {
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 %1, i32* %.threadid_temp., align 4 store i32 %1, i32* %.threadid_temp., align 4
call void @__omp_outlined__6(i32* %.threadid_temp., i32* %.zero.addr) #3 call void @__omp_outlined__6(i32* %.threadid_temp., i32* %.zero.addr) #3
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
define internal void @__omp_outlined__6(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__6(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
entry: entry:
▲ Show 20 Lines • Show All 63 Lines • ▼ Show 20 Linesentry:
ret void ret void
} }
define weak void @__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66() #0 { define weak void @__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66() #0 {
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 %1, i32* %.threadid_temp., align 4 store i32 %1, i32* %.threadid_temp., align 4
call void @__omp_outlined__9(i32* %.threadid_temp., i32* %.zero.addr) #3 call void @__omp_outlined__9(i32* %.threadid_temp., i32* %.zero.addr) #3
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
define internal void @__omp_outlined__9(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__9(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
entry: entry:
▲ Show 20 Lines • Show All 61 Lines • ▼ Show 20 Linesentry:
ret void ret void
} }
define weak void @__omp_offloading_14_a36502b_simple_state_machine_pure_l77() #0 { define weak void @__omp_offloading_14_a36502b_simple_state_machine_pure_l77() #0 {
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 %1, i32* %.threadid_temp., align 4 store i32 %1, i32* %.threadid_temp., align 4
call void @__omp_outlined__12(i32* %.threadid_temp., i32* %.zero.addr) #3 call void @__omp_outlined__12(i32* %.threadid_temp., i32* %.zero.addr) #3
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
define internal void @__omp_outlined__12(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__12(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
entry: entry:
▲ Show 20 Lines • Show All 64 Lines • ▼ Show 20 Linesentry:
ret void ret void
} }
define weak void @__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92() #0 { define weak void @__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92() #0 {
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 %1, i32* %.threadid_temp., align 4 store i32 %1, i32* %.threadid_temp., align 4
call void @__omp_outlined__15(i32* %.threadid_temp., i32* %.zero.addr) #3 call void @__omp_outlined__15(i32* %.threadid_temp., i32* %.zero.addr) #3
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
define internal void @__omp_outlined__15(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__15(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
entry: entry:
Show All 30 Lines
declare i32 @omp_get_thread_num(...) #4 declare i32 @omp_get_thread_num(...) #4
define weak void @__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112() #0 { define weak void @__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112() #0 {
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 %1, i32* %.threadid_temp., align 4 store i32 %1, i32* %.threadid_temp., align 4
call void @__omp_outlined__16(i32* %.threadid_temp., i32* %.zero.addr) #3 call void @__omp_outlined__16(i32* %.threadid_temp., i32* %.zero.addr) #3
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
define internal void @__omp_outlined__16(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__16(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
entry: entry:
▲ Show 20 Lines • Show All 132 Lines • ▼ Show 20 Lines
!17 = !{i32 7, !"openmp", i32 50} !17 = !{i32 7, !"openmp", i32 50}
!18 = !{i32 7, !"openmp-device", i32 50} !18 = !{i32 7, !"openmp-device", i32 50}
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_needed_l14 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_needed_l14
; AMDGPU-SAME: () #[[ATTR0:[0-9]+]] { ; AMDGPU-SAME: () #[[ATTR0:[0-9]+]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 1, i1 true)
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3:[0-9]+]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3:[0-9]+]]
; AMDGPU-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: worker.exit: ; AMDGPU: worker.exit:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__ ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
Show All 40 Lines
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_l22 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_l22
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; AMDGPU-NEXT: call void @__omp_outlined__1(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-NEXT: call void @__omp_outlined__1(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: worker.exit: ; AMDGPU: worker.exit:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__1 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__1
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 61 Lines • ▼ Show 20 Lines
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; AMDGPU-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: worker.exit: ; AMDGPU: worker.exit:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__4 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__4
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 81 Lines • ▼ Show 20 Lines
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; AMDGPU-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: worker.exit: ; AMDGPU: worker.exit:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__6 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__6
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; AMDGPU-NEXT: call void @__omp_outlined__9(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-NEXT: call void @__omp_outlined__9(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: worker.exit: ; AMDGPU: worker.exit:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__9 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__9
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_pure_l77 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_pure_l77
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; AMDGPU-NEXT: call void @__omp_outlined__12(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-NEXT: call void @__omp_outlined__12(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: worker.exit: ; AMDGPU: worker.exit:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__12 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__12
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__omp_outlined__15(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-NEXT: call void @__omp_outlined__15(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: worker.exit: ; AMDGPU: worker.exit:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__15 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__15
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__omp_outlined__16(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; AMDGPU-NEXT: call void @__omp_outlined__16(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: worker.exit: ; AMDGPU: worker.exit:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; ;
; ;
; AMDGPU: Function Attrs: convergent noinline norecurse nounwind ; AMDGPU: Function Attrs: convergent noinline norecurse nounwind
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__16 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__16
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 103 Lines • ▼ Show 20 Lines
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_needed_l14 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_needed_l14
; NVPTX-SAME: () #[[ATTR0:[0-9]+]] { ; NVPTX-SAME: () #[[ATTR0:[0-9]+]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 1, i1 true, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 1, i1 true)
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3:[0-9]+]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3:[0-9]+]]
; NVPTX-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: worker.exit: ; NVPTX: worker.exit:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__ ; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__
; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
Show All 40 Lines
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_l22 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_l22
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; NVPTX-NEXT: call void @__omp_outlined__1(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-NEXT: call void @__omp_outlined__1(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: worker.exit: ; NVPTX: worker.exit:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__1 ; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__1
; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 61 Lines • ▼ Show 20 Lines
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_l39
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; NVPTX-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: worker.exit: ; NVPTX: worker.exit:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__4 ; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__4
; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 81 Lines • ▼ Show 20 Lines
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_with_fallback_l55
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; NVPTX-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: worker.exit: ; NVPTX: worker.exit:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__6 ; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__6
; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_no_openmp_attr_l66
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; NVPTX-NEXT: call void @__omp_outlined__9(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-NEXT: call void @__omp_outlined__9(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: worker.exit: ; NVPTX: worker.exit:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__9 ; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__9
; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_pure_l77 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_pure_l77
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4 ; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4
; NVPTX-NEXT: call void @__omp_outlined__12(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-NEXT: call void @__omp_outlined__12(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: worker.exit: ; NVPTX: worker.exit:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__12 ; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__12
; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_simple_state_machine_interprocedural_nested_recursive_l92
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__omp_outlined__15(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-NEXT: call void @__omp_outlined__15(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: worker.exit: ; NVPTX: worker.exit:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__15 ; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__15
; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 48 Lines • ▼ Show 20 Lines
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_14_a36502b_no_state_machine_weak_callee_l112
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 true)
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__omp_outlined__16(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]] ; NVPTX-NEXT: call void @__omp_outlined__16(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR3]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: worker.exit: ; NVPTX: worker.exit:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; ;
; ;
; NVPTX: Function Attrs: convergent noinline norecurse nounwind ; NVPTX: Function Attrs: convergent noinline norecurse nounwind
; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__16 ; NVPTX-LABEL: define {{[^@]+}}@__omp_outlined__16
; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; NVPTX-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
▲ Show 20 Lines • Show All 104 Lines • Show Last 20 Lines

llvm/test/Transforms/OpenMP/custom_state_machines_remarks.ll

Show First 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
@13 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 2, i32 0, i8* getelementptr inbounds ([73 x i8], [73 x i8]* @12, i32 0, i32 0) }, align 8 @13 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 2, i32 0, i8* getelementptr inbounds ([73 x i8], [73 x i8]* @12, i32 0, i32 0) }, align 8
@G = external global i32 @G = external global i32
@llvm.compiler.used = appending global [2 x i8*] [i8* @__omp_offloading_2a_d80d3d_test_fallback_l11_exec_mode, i8* @__omp_offloading_2a_d80d3d_test_no_fallback_l20_exec_mode], section "llvm.metadata" @llvm.compiler.used = appending global [2 x i8*] [i8* @__omp_offloading_2a_d80d3d_test_fallback_l11_exec_mode, i8* @__omp_offloading_2a_d80d3d_test_no_fallback_l20_exec_mode], section "llvm.metadata"
; Function Attrs: convergent norecurse nounwind ; Function Attrs: convergent norecurse nounwind
define weak void @__omp_offloading_2a_d80d3d_test_fallback_l11() local_unnamed_addr #0 !dbg !15 { define weak void @__omp_offloading_2a_d80d3d_test_fallback_l11() local_unnamed_addr #0 !dbg !15 {
entry: entry:
%captured_vars_addrs.i.i = alloca [0 x i8*], align 8 %captured_vars_addrs.i.i = alloca [0 x i8*], align 8
%0 = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @1, i8 1, i1 true, i1 true) #3, !dbg !18 %0 = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @1, i8 1, i1 true) #3, !dbg !18
%exec_user_code = icmp eq i32 %0, -1, !dbg !18 %exec_user_code = icmp eq i32 %0, -1, !dbg !18
br i1 %exec_user_code, label %user_code.entry, label %common.ret, !dbg !18 br i1 %exec_user_code, label %user_code.entry, label %common.ret, !dbg !18
common.ret: ; preds = %entry, %user_code.entry common.ret: ; preds = %entry, %user_code.entry
ret void, !dbg !19 ret void, !dbg !19
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @3) #3 %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @3) #3
call void @unknown() #6, !dbg !20 call void @unknown() #6, !dbg !20
%2 = bitcast [0 x i8*]* %captured_vars_addrs.i.i to i8* %2 = bitcast [0 x i8*]* %captured_vars_addrs.i.i to i8*
call void @llvm.lifetime.start.p0i8(i64 0, i8* nonnull %2) #3 call void @llvm.lifetime.start.p0i8(i64 0, i8* nonnull %2) #3
%3 = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @13) #3 %3 = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @13) #3
%4 = getelementptr inbounds [0 x i8*], [0 x i8*]* %captured_vars_addrs.i.i, i64 0, i64 0, !dbg !23 %4 = getelementptr inbounds [0 x i8*], [0 x i8*]* %captured_vars_addrs.i.i, i64 0, i64 0, !dbg !23
call void @__kmpc_parallel_51(%struct.ident_t* noundef nonnull @13, i32 %3, i32 noundef 1, i32 noundef -1, i32 noundef -1, i8* noundef bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* noundef bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** noundef nonnull %4, i64 noundef 0) #3, !dbg !23 call void @__kmpc_parallel_51(%struct.ident_t* noundef nonnull @13, i32 %3, i32 noundef 1, i32 noundef -1, i32 noundef -1, i8* noundef bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* noundef bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** noundef nonnull %4, i64 noundef 0) #3, !dbg !23
call void @llvm.lifetime.end.p0i8(i64 0, i8* nonnull %2) #3, !dbg !26 call void @llvm.lifetime.end.p0i8(i64 0, i8* nonnull %2) #3, !dbg !26
call void @unknown() #6, !dbg !27 call void @unknown() #6, !dbg !27
call void @__kmpc_target_deinit(%struct.ident_t* nonnull @5, i8 1, i1 true) #3, !dbg !28 call void @__kmpc_target_deinit(%struct.ident_t* nonnull @5, i8 1) #3, !dbg !28
br label %common.ret br label %common.ret
} }
; Make it a weak definition so we will apply custom state machine rewriting but can't use the body in the reasoning. ; Make it a weak definition so we will apply custom state machine rewriting but can't use the body in the reasoning.
define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1, i1) { define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1) {
ret i32 0 ret i32 0
} }
; Function Attrs: convergent ; Function Attrs: convergent
declare void @unknown() local_unnamed_addr #1 declare void @unknown() local_unnamed_addr #1
; Function Attrs: nounwind ; Function Attrs: nounwind
define hidden void @known() local_unnamed_addr #2 !dbg !29 { define hidden void @known() local_unnamed_addr #2 !dbg !29 {
entry: entry:
%captured_vars_addrs = alloca [0 x i8*], align 8 %captured_vars_addrs = alloca [0 x i8*], align 8
%0 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @13) %0 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @13)
%1 = getelementptr inbounds [0 x i8*], [0 x i8*]* %captured_vars_addrs, i64 0, i64 0, !dbg !30 %1 = getelementptr inbounds [0 x i8*], [0 x i8*]* %captured_vars_addrs, i64 0, i64 0, !dbg !30
call void @__kmpc_parallel_51(%struct.ident_t* nonnull @13, i32 %0, i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** nonnull %1, i64 0) #3, !dbg !30 call void @__kmpc_parallel_51(%struct.ident_t* nonnull @13, i32 %0, i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** nonnull %1, i64 0) #3, !dbg !30
ret void, !dbg !31 ret void, !dbg !31
} }
; Function Attrs: nounwind ; Function Attrs: nounwind
declare i32 @__kmpc_global_thread_num(%struct.ident_t*) local_unnamed_addr #3 declare i32 @__kmpc_global_thread_num(%struct.ident_t*) local_unnamed_addr #3
declare void @__kmpc_target_deinit(%struct.ident_t*, i8, i1) local_unnamed_addr declare void @__kmpc_target_deinit(%struct.ident_t*, i8) local_unnamed_addr
; Function Attrs: norecurse nounwind ; Function Attrs: norecurse nounwind
define weak void @__omp_offloading_2a_d80d3d_test_no_fallback_l20() local_unnamed_addr #4 !dbg !32 { define weak void @__omp_offloading_2a_d80d3d_test_no_fallback_l20() local_unnamed_addr #4 !dbg !32 {
entry: entry:
%captured_vars_addrs.i2.i = alloca [0 x i8*], align 8 %captured_vars_addrs.i2.i = alloca [0 x i8*], align 8
%0 = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @7, i8 1, i1 true, i1 true) #3, !dbg !33 %0 = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @7, i8 1, i1 true) #3, !dbg !33
%exec_user_code = icmp eq i32 %0, -1, !dbg !33 %exec_user_code = icmp eq i32 %0, -1, !dbg !33
br i1 %exec_user_code, label %user_code.entry, label %common.ret, !dbg !33 br i1 %exec_user_code, label %user_code.entry, label %common.ret, !dbg !33
common.ret: ; preds = %entry, %user_code.entry common.ret: ; preds = %entry, %user_code.entry
ret void, !dbg !34 ret void, !dbg !34
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @9) #3 %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @9) #3
%2 = bitcast [0 x i8*]* %captured_vars_addrs.i2.i to i8* %2 = bitcast [0 x i8*]* %captured_vars_addrs.i2.i to i8*
call void @llvm.lifetime.start.p0i8(i64 0, i8* nonnull %2) #3 call void @llvm.lifetime.start.p0i8(i64 0, i8* nonnull %2) #3
%3 = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @13) #3 %3 = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @13) #3
%4 = getelementptr inbounds [0 x i8*], [0 x i8*]* %captured_vars_addrs.i2.i, i64 0, i64 0, !dbg !35 %4 = getelementptr inbounds [0 x i8*], [0 x i8*]* %captured_vars_addrs.i2.i, i64 0, i64 0, !dbg !35
call void @__kmpc_parallel_51(%struct.ident_t* noundef nonnull @13, i32 %3, i32 noundef 1, i32 noundef -1, i32 noundef -1, i8* noundef bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* noundef bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** noundef nonnull %4, i64 noundef 0) #3, !dbg !35 call void @__kmpc_parallel_51(%struct.ident_t* noundef nonnull @13, i32 %3, i32 noundef 1, i32 noundef -1, i32 noundef -1, i8* noundef bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* noundef bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** noundef nonnull %4, i64 noundef 0) #3, !dbg !35
call void @llvm.lifetime.end.p0i8(i64 0, i8* nonnull %2) #3, !dbg !39 call void @llvm.lifetime.end.p0i8(i64 0, i8* nonnull %2) #3, !dbg !39
call void @llvm.lifetime.start.p0i8(i64 0, i8* nonnull %2) #3 call void @llvm.lifetime.start.p0i8(i64 0, i8* nonnull %2) #3
%5 = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @13) #3 %5 = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @13) #3
call void @__kmpc_parallel_51(%struct.ident_t* noundef nonnull @13, i32 %5, i32 noundef 1, i32 noundef -1, i32 noundef -1, i8* noundef bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* noundef bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** noundef nonnull %4, i64 noundef 0) #3, !dbg !40 call void @__kmpc_parallel_51(%struct.ident_t* noundef nonnull @13, i32 %5, i32 noundef 1, i32 noundef -1, i32 noundef -1, i8* noundef bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* noundef bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** noundef nonnull %4, i64 noundef 0) #3, !dbg !40
call void @llvm.lifetime.end.p0i8(i64 0, i8* nonnull %2) #3, !dbg !42 call void @llvm.lifetime.end.p0i8(i64 0, i8* nonnull %2) #3, !dbg !42
call void @llvm.lifetime.start.p0i8(i64 0, i8* nonnull %2) #3 call void @llvm.lifetime.start.p0i8(i64 0, i8* nonnull %2) #3
%6 = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @13) #3 %6 = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @13) #3
call void @__kmpc_parallel_51(%struct.ident_t* noundef nonnull @13, i32 %6, i32 noundef 1, i32 noundef -1, i32 noundef -1, i8* noundef bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* noundef bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** noundef nonnull %4, i64 noundef 0) #3, !dbg !43 call void @__kmpc_parallel_51(%struct.ident_t* noundef nonnull @13, i32 %6, i32 noundef 1, i32 noundef -1, i32 noundef -1, i8* noundef bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* noundef bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** noundef nonnull %4, i64 noundef 0) #3, !dbg !43
call void @llvm.lifetime.end.p0i8(i64 0, i8* nonnull %2) #3, !dbg !45 call void @llvm.lifetime.end.p0i8(i64 0, i8* nonnull %2) #3, !dbg !45
call void @no_openmp() call void @no_openmp()
call void @no_parallelism() call void @no_parallelism()
call void @__kmpc_target_deinit(%struct.ident_t* nonnull @11, i8 1, i1 true) #3, !dbg !46 call void @__kmpc_target_deinit(%struct.ident_t* nonnull @11, i8 1) #3, !dbg !46
br label %common.ret br label %common.ret
} }
; Function Attrs: convergent norecurse nounwind ; Function Attrs: convergent norecurse nounwind
define internal void @__omp_outlined__2(i32* noalias nocapture nofree readnone %.global_tid., i32* noalias nocapture nofree readnone %.bound_tid.) #0 !dbg !47 { define internal void @__omp_outlined__2(i32* noalias nocapture nofree readnone %.global_tid., i32* noalias nocapture nofree readnone %.bound_tid.) #0 !dbg !47 {
entry: entry:
call void @unknown() #6, !dbg !48 call void @unknown() #6, !dbg !48
ret void, !dbg !49 ret void, !dbg !49
▲ Show 20 Lines • Show All 94 Lines • Show Last 20 Lines

llvm/test/Transforms/OpenMP/deduplication_target.ll

Show All 13 Lines
declare void @use(i32) declare void @use(i32)
define weak void @__omp_offloading_50_a3e09bf8_foo_l2() #0 { define weak void @__omp_offloading_50_a3e09bf8_foo_l2() #0 {
; CHECK-LABEL: define {{[^@]+}}@__omp_offloading_50_a3e09bf8_foo_l2 ; CHECK-LABEL: define {{[^@]+}}@__omp_offloading_50_a3e09bf8_foo_l2
; CHECK-SAME: () #[[ATTR0:[0-9]+]] { ; CHECK-SAME: () #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x i8*], align 8 ; CHECK-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x i8*], align 8
; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 true) ; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1:[0-9]+]], i8 2, i1 false)
; CHECK-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]]) ; CHECK-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2:[0-9]+]])
; CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; 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-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; CHECK: user_code.entry: ; CHECK: user_code.entry:
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 true) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; CHECK: worker.exit: ; CHECK: worker.exit:
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
entry: entry:
%captured_vars_addrs = alloca [0 x i8*], align 8 %captured_vars_addrs = alloca [0 x i8*], align 8
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 2, i1 false, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 2, i1 false)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @2) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @2)
%2 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @2) %2 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @2)
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 2, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 2)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1, i1) declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1)
declare i32 @__kmpc_global_thread_num(%struct.ident_t*) #1 declare i32 @__kmpc_global_thread_num(%struct.ident_t*) #1
declare void @__kmpc_target_deinit(%struct.ident_t*, i8, i1) declare void @__kmpc_target_deinit(%struct.ident_t*, i8)
attributes #0 = { convergent noinline norecurse nounwind "frame-pointer"="none" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-features"="+ptx32,+sm_20" } attributes #0 = { convergent noinline norecurse nounwind "frame-pointer"="none" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-features"="+ptx32,+sm_20" }
attributes #1 = { nounwind } attributes #1 = { nounwind }
!omp_offload.info = !{!0} !omp_offload.info = !{!0}
!nvvm.annotations = !{!1} !nvvm.annotations = !{!1}
!llvm.module.flags = !{!2, !3, !4} !llvm.module.flags = !{!2, !3, !4}
!0 = !{i32 0, i32 80, i32 -1545561096, !"foo", i32 2, i32 0} !0 = !{i32 0, i32 80, i32 -1545561096, !"foo", i32 2, i32 0}
!1 = !{void ()* @__omp_offloading_50_a3e09bf8_foo_l2, !"kernel", i32 1} !1 = !{void ()* @__omp_offloading_50_a3e09bf8_foo_l2, !"kernel", i32 1}
!2 = !{i32 1, !"wchar_size", i32 4} !2 = !{i32 1, !"wchar_size", i32 4}
!3 = !{i32 7, !"openmp", i32 50} !3 = !{i32 7, !"openmp", i32 50}
!4 = !{i32 7, !"openmp-device", i32 50} !4 = !{i32 7, !"openmp-device", i32 50}

llvm/test/Transforms/OpenMP/fold_generic_main_thread.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature
; RUN: opt -S -passes='openmp-opt' < %s | FileCheck %s ; RUN: opt -S -passes='openmp-opt' < %s | FileCheck %s
; RUN: opt -openmp-opt-disable-folding -S -passes='openmp-opt' < %s | FileCheck %s --check-prefix=CHECK-DISABLED ; RUN: opt -openmp-opt-disable-folding -S -passes='openmp-opt' < %s | FileCheck %s --check-prefix=CHECK-DISABLED
; ModuleID = 'single_threaded_exeuction.c' ; ModuleID = 'single_threaded_exeuction.c'
%struct.ident_t = type { i32, i32, i32, i32, i8* } %struct.ident_t = type { i32, i32, i32, i32, i8* }
@0 = private unnamed_addr constant [1 x i8] c"\00", align 1 @0 = private unnamed_addr constant [1 x i8] c"\00", align 1
@1 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([1 x i8], [1 x i8]* @0, i32 0, i32 0) }, align 8 @1 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([1 x i8], [1 x i8]* @0, i32 0, i32 0) }, align 8
define void @kernel() { define void @kernel() {
; CHECK-LABEL: define {{[^@]+}}@kernel() { ; CHECK-LABEL: define {{[^@]+}}@kernel() {
; CHECK-NEXT: [[CALL:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @[[GLOB1:[0-9]+]], i8 1, i1 false, i1 false) ; CHECK-NEXT: [[CALL:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @[[GLOB1:[0-9]+]], i8 1, i1 false)
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[CALL]], -1 ; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[CALL]], -1
; CHECK-NEXT: br i1 [[CMP]], label [[IF_THEN:%.*]], label [[IF_ELSE:%.*]] ; CHECK-NEXT: br i1 [[CMP]], label [[IF_THEN:%.*]], label [[IF_ELSE:%.*]]
; CHECK: if.then: ; CHECK: if.then:
; CHECK-NEXT: call void @foo() ; CHECK-NEXT: call void @foo()
; CHECK-NEXT: br label [[IF_END:%.*]] ; CHECK-NEXT: br label [[IF_END:%.*]]
; CHECK: if.else: ; CHECK: if.else:
; CHECK-NEXT: call void @bar() ; CHECK-NEXT: call void @bar()
; CHECK-NEXT: br label [[IF_END]] ; CHECK-NEXT: br label [[IF_END]]
; CHECK: if.end: ; CHECK: if.end:
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1, i1 true) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
; CHECK-DISABLED-LABEL: define {{[^@]+}}@kernel() { ; CHECK-DISABLED-LABEL: define {{[^@]+}}@kernel() {
; CHECK-DISABLED-NEXT: [[CALL:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @[[GLOB1:[0-9]+]], i8 1, i1 false, i1 false) ; CHECK-DISABLED-NEXT: [[CALL:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @[[GLOB1:[0-9]+]], i8 1, i1 false)
; CHECK-DISABLED-NEXT: [[CMP:%.*]] = icmp eq i32 [[CALL]], -1 ; CHECK-DISABLED-NEXT: [[CMP:%.*]] = icmp eq i32 [[CALL]], -1
; CHECK-DISABLED-NEXT: br i1 [[CMP]], label [[IF_THEN:%.*]], label [[IF_ELSE:%.*]] ; CHECK-DISABLED-NEXT: br i1 [[CMP]], label [[IF_THEN:%.*]], label [[IF_ELSE:%.*]]
; CHECK-DISABLED: if.then: ; CHECK-DISABLED: if.then:
; CHECK-DISABLED-NEXT: call void @foo() ; CHECK-DISABLED-NEXT: call void @foo()
; CHECK-DISABLED-NEXT: br label [[IF_END:%.*]] ; CHECK-DISABLED-NEXT: br label [[IF_END:%.*]]
; CHECK-DISABLED: if.else: ; CHECK-DISABLED: if.else:
; CHECK-DISABLED-NEXT: call void @bar() ; CHECK-DISABLED-NEXT: call void @bar()
; CHECK-DISABLED-NEXT: br label [[IF_END]] ; CHECK-DISABLED-NEXT: br label [[IF_END]]
; CHECK-DISABLED: if.end: ; CHECK-DISABLED: if.end:
; CHECK-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1, i1 true) ; CHECK-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1)
; CHECK-DISABLED-NEXT: ret void ; CHECK-DISABLED-NEXT: ret void
; ;
%call = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @1, i8 1, i1 false, i1 false) %call = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @1, i8 1, i1 false)
%cmp = icmp eq i32 %call, -1 %cmp = icmp eq i32 %call, -1
br i1 %cmp, label %if.then, label %if.else br i1 %cmp, label %if.then, label %if.else
if.then: if.then:
call void @foo() call void @foo()
br label %if.end br label %if.end
if.else: if.else:
call void @bar() call void @bar()
br label %if.end br label %if.end
if.end: if.end:
call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1)
ret void ret void
} }
; Function Attrs: noinline ; Function Attrs: noinline
define internal void @foo() { define internal void @foo() {
; CHECK-LABEL: define {{[^@]+}}@foo() { ; CHECK-LABEL: define {{[^@]+}}@foo() {
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: [[TID:%.*]] = call i32 @__kmpc_get_hardware_thread_id() ; CHECK-NEXT: [[TID:%.*]] = call i32 @__kmpc_get_hardware_thread_id()
▲ Show 20 Lines • Show All 71 Lines • ▼ Show 20 Lines
} }
declare void @baz() declare void @baz()
declare i8 @__kmpc_is_generic_main_thread_id(i32) declare i8 @__kmpc_is_generic_main_thread_id(i32)
declare i32 @__kmpc_get_hardware_thread_id() declare i32 @__kmpc_get_hardware_thread_id()
declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1, i1) declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1)
declare void @__kmpc_target_deinit(%struct.ident_t*, i8, i1) declare void @__kmpc_target_deinit(%struct.ident_t*, i8)
!llvm.dbg.cu = !{!0} !llvm.dbg.cu = !{!0}
!llvm.module.flags = !{!3, !4, !5, !6} !llvm.module.flags = !{!3, !4, !5, !6}
!nvvm.annotations = !{!7} !nvvm.annotations = !{!7}
!0 = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 13.0.0", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, enums: !2, splitDebugInlining: false, nameTableKind: None) !0 = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 13.0.0", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, enums: !2, splitDebugInlining: false, nameTableKind: None)
!1 = !DIFile(filename: "fold_generic_main_thread.c", directory: "/tmp/fold_generic_main_thread.c") !1 = !DIFile(filename: "fold_generic_main_thread.c", directory: "/tmp/fold_generic_main_thread.c")
!2 = !{} !2 = !{}
!3 = !{i32 2, !"Debug Info Version", i32 3} !3 = !{i32 2, !"Debug Info Version", i32 3}
!4 = !{i32 1, !"wchar_size", i32 4} !4 = !{i32 1, !"wchar_size", i32 4}
!5 = !{i32 7, !"openmp", i32 50} !5 = !{i32 7, !"openmp", i32 50}
!6 = !{i32 7, !"openmp-device", i32 50} !6 = !{i32 7, !"openmp-device", i32 50}
!7 = !{void ()* @kernel, !"kernel", i32 1} !7 = !{void ()* @kernel, !"kernel", i32 1}

llvm/test/Transforms/OpenMP/get_hardware_num_threads_in_block_fold.ll

Show All 12 Lines
; CHECK: @[[KERNEL1_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 3 ; CHECK: @[[KERNEL1_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 3
; CHECK: @[[KERNEL2_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 3 ; CHECK: @[[KERNEL2_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 3
; CHECK: @[[GLOB0:[0-9]+]] = private unnamed_addr constant [23 x i8] c" ; CHECK: @[[GLOB0:[0-9]+]] = private unnamed_addr constant [23 x i8] c"
; CHECK: @[[GLOB1:[0-9]+]] = private unnamed_addr constant [[STRUCT_IDENT_T:%.*]] { i32 0, i32 2, i32 0, i32 22, i8* getelementptr inbounds ([23 x i8], [23 x i8]* @[[GLOB0]], i32 0, i32 0) }, align 8 ; CHECK: @[[GLOB1:[0-9]+]] = private unnamed_addr constant [[STRUCT_IDENT_T:%.*]] { i32 0, i32 2, i32 0, i32 22, i8* getelementptr inbounds ([23 x i8], [23 x i8]* @[[GLOB0]], i32 0, i32 0) }, align 8
;. ;.
define weak void @kernel0() #0 { define weak void @kernel0() #0 {
; CHECK-LABEL: define {{[^@]+}}@kernel0 ; CHECK-LABEL: define {{[^@]+}}@kernel0
; CHECK-SAME: () #[[ATTR0:[0-9]+]] { ; CHECK-SAME: () #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false, i1 false) ; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false)
; CHECK-NEXT: [[THREAD_ID_IN_BLOCK:%.*]] = call i32 @__kmpc_get_hardware_thread_id_in_block() ; CHECK-NEXT: [[THREAD_ID_IN_BLOCK:%.*]] = call i32 @__kmpc_get_hardware_thread_id_in_block()
; CHECK-NEXT: [[THREAD_IS_MAIN:%.*]] = icmp ne i32 [[THREAD_ID_IN_BLOCK]], 0 ; CHECK-NEXT: [[THREAD_IS_MAIN:%.*]] = icmp ne i32 [[THREAD_ID_IN_BLOCK]], 0
; CHECK-NEXT: br i1 [[THREAD_IS_MAIN]], label [[EXIT_THREADS:%.*]], label [[MAIN_THREAD_USER_CODE:%.*]] ; CHECK-NEXT: br i1 [[THREAD_IS_MAIN]], label [[EXIT_THREADS:%.*]], label [[MAIN_THREAD_USER_CODE:%.*]]
; CHECK: exit.threads: ; CHECK: exit.threads:
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; CHECK: main.thread.user_code: ; CHECK: main.thread.user_code:
; CHECK-NEXT: call void @helper0() #[[ATTR1:[0-9]+]] ; CHECK-NEXT: call void @helper0() #[[ATTR1:[0-9]+]]
; CHECK-NEXT: call void @helper1() #[[ATTR1]] ; CHECK-NEXT: call void @helper1() #[[ATTR1]]
; CHECK-NEXT: call void @helper2() #[[ATTR1]] ; CHECK-NEXT: call void @helper2() #[[ATTR1]]
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2, i1 false) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
%i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 false, i1 false) %i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 false)
call void @helper0() call void @helper0()
call void @helper1() call void @helper1()
call void @helper2() call void @helper2()
call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1, i1 false) call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1)
ret void ret void
} }
@kernel1_exec_mode = weak constant i8 1 @kernel1_exec_mode = weak constant i8 1
define weak void @kernel1() #0 { define weak void @kernel1() #0 {
; CHECK-LABEL: define {{[^@]+}}@kernel1 ; CHECK-LABEL: define {{[^@]+}}@kernel1
; CHECK-SAME: () #[[ATTR0]] { ; CHECK-SAME: () #[[ATTR0]] {
; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false, i1 false) ; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false)
; CHECK-NEXT: [[THREAD_ID_IN_BLOCK:%.*]] = call i32 @__kmpc_get_hardware_thread_id_in_block() ; CHECK-NEXT: [[THREAD_ID_IN_BLOCK:%.*]] = call i32 @__kmpc_get_hardware_thread_id_in_block()
; CHECK-NEXT: [[THREAD_IS_MAIN:%.*]] = icmp ne i32 [[THREAD_ID_IN_BLOCK]], 0 ; CHECK-NEXT: [[THREAD_IS_MAIN:%.*]] = icmp ne i32 [[THREAD_ID_IN_BLOCK]], 0
; CHECK-NEXT: br i1 [[THREAD_IS_MAIN]], label [[EXIT_THREADS:%.*]], label [[MAIN_THREAD_USER_CODE:%.*]] ; CHECK-NEXT: br i1 [[THREAD_IS_MAIN]], label [[EXIT_THREADS:%.*]], label [[MAIN_THREAD_USER_CODE:%.*]]
; CHECK: exit.threads: ; CHECK: exit.threads:
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; CHECK: main.thread.user_code: ; CHECK: main.thread.user_code:
; CHECK-NEXT: call void @helper1() #[[ATTR1]] ; CHECK-NEXT: call void @helper1() #[[ATTR1]]
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2, i1 false) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
%i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 false, i1 false) %i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 false)
call void @helper1() call void @helper1()
call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1, i1 false) call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1)
ret void ret void
} }
@kernel2_exec_mode = weak constant i8 1 @kernel2_exec_mode = weak constant i8 1
define weak void @kernel2() #0 { define weak void @kernel2() #0 {
; CHECK-LABEL: define {{[^@]+}}@kernel2 ; CHECK-LABEL: define {{[^@]+}}@kernel2
; CHECK-SAME: () #[[ATTR0]] { ; CHECK-SAME: () #[[ATTR0]] {
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x i8*], align 8 ; CHECK-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x i8*], align 8
; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false, i1 false) ; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false)
; CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[I]], -1 ; CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[I]], -1
; CHECK-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; CHECK-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; CHECK: common.ret: ; CHECK: common.ret:
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; CHECK: user_code.entry: ; CHECK: user_code.entry:
; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* null) #[[ATTR1]] ; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* null) #[[ATTR1]]
; CHECK-NEXT: [[TMP1:%.*]] = bitcast [0 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8** ; CHECK-NEXT: [[TMP1:%.*]] = bitcast [0 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8**
; CHECK-NEXT: call void @helper0() #[[ATTR1]] ; CHECK-NEXT: call void @helper0() #[[ATTR1]]
; CHECK-NEXT: call void @helper1() #[[ATTR1]] ; CHECK-NEXT: call void @helper1() #[[ATTR1]]
; CHECK-NEXT: call void @helper2() #[[ATTR1]] ; CHECK-NEXT: call void @helper2() #[[ATTR1]]
; CHECK-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* null, i32 [[TMP0]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__ to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined___wrapper to i8*), i8** [[TMP1]], i64 0) ; CHECK-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* null, i32 [[TMP0]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__ to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined___wrapper to i8*), i8** [[TMP1]], i64 0)
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2, i1 false) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
entry: entry:
%captured_vars_addrs = alloca [0 x i8*], align 8 %captured_vars_addrs = alloca [0 x i8*], align 8
%i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 true, i1 true) %i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 true)
%exec_user_code = icmp eq i32 %i, -1 %exec_user_code = icmp eq i32 %i, -1
br i1 %exec_user_code, label %user_code.entry, label %common.ret br i1 %exec_user_code, label %user_code.entry, label %common.ret
common.ret: common.ret:
ret void ret void
user_code.entry: user_code.entry:
%0 = call i32 @__kmpc_global_thread_num(%struct.ident_t* null) %0 = call i32 @__kmpc_global_thread_num(%struct.ident_t* null)
%1 = bitcast [0 x i8*]* %captured_vars_addrs to i8** %1 = bitcast [0 x i8*]* %captured_vars_addrs to i8**
call void @helper0() call void @helper0()
call void @helper1() call void @helper1()
call void @helper2() call void @helper2()
call void @__kmpc_parallel_51(%struct.ident_t* null, i32 %0, i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__ to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined___wrapper to i8*), i8** %1, i64 0) call void @__kmpc_parallel_51(%struct.ident_t* null, i32 %0, i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__ to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined___wrapper to i8*), i8** %1, i64 0)
call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1)
ret void ret void
} }
define internal void @helper0() { define internal void @helper0() {
; CHECK-LABEL: define {{[^@]+}}@helper0 ; CHECK-LABEL: define {{[^@]+}}@helper0
; CHECK-SAME: () #[[ATTR1]] { ; CHECK-SAME: () #[[ATTR1]] {
; CHECK-NEXT: br label [[REGION_CHECK_TID:%.*]] ; CHECK-NEXT: br label [[REGION_CHECK_TID:%.*]]
; CHECK: region.check.tid: ; CHECK: region.check.tid:
▲ Show 20 Lines • Show All 84 Lines • ▼ Show 20 Lines
; CHECK-SAME: () #[[ATTR1]] { ; CHECK-SAME: () #[[ATTR1]] {
; CHECK-NEXT: [[RET:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block_dummy() ; CHECK-NEXT: [[RET:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block_dummy()
; CHECK-NEXT: ret i32 [[RET]] ; CHECK-NEXT: ret i32 [[RET]]
; ;
%ret = call i32 @__kmpc_get_hardware_num_threads_in_block_dummy() %ret = call i32 @__kmpc_get_hardware_num_threads_in_block_dummy()
ret i32 %ret ret i32 %ret
} }
declare i32 @__kmpc_get_hardware_num_threads_in_block_dummy() declare i32 @__kmpc_get_hardware_num_threads_in_block_dummy()
declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1 zeroext, i1 zeroext) #1 declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1 zeroext) #1
declare void @__kmpc_target_deinit(%struct.ident_t* nocapture readnone, i8, i1 zeroext) #1 declare void @__kmpc_target_deinit(%struct.ident_t* nocapture readnone, i8) #1
declare void @__kmpc_parallel_51(%struct.ident_t*, i32, i32, i32, i32, i8*, i8*, i8**, i64) declare void @__kmpc_parallel_51(%struct.ident_t*, i32, i32, i32, i32, i8*, i8*, i8**, i64)
declare i32 @__kmpc_global_thread_num(%struct.ident_t*) declare i32 @__kmpc_global_thread_num(%struct.ident_t*)
!llvm.module.flags = !{!0, !1} !llvm.module.flags = !{!0, !1}
!nvvm.annotations = !{!2, !3, !4} !nvvm.annotations = !{!2, !3, !4}
attributes #0 = { "omp_target_thread_limit"="666" "omp_target_num_teams"="777"} attributes #0 = { "omp_target_thread_limit"="666" "omp_target_num_teams"="777"}
Show All 19 Lines

llvm/test/Transforms/OpenMP/get_hardware_num_threads_in_block_fold_optnone.ll

Show All 10 Lines
; CHECK: @[[KERNEL0_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 1 ; CHECK: @[[KERNEL0_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 1
; CHECK: @[[G:[a-zA-Z0-9_$"\\.-]+]] = external global i32 ; CHECK: @[[G:[a-zA-Z0-9_$"\\.-]+]] = external global i32
; CHECK: @[[KERNEL1_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 1 ; CHECK: @[[KERNEL1_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 1
; CHECK: @[[KERNEL2_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 1 ; CHECK: @[[KERNEL2_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 1
;. ;.
define weak void @kernel0() #0 { define weak void @kernel0() #0 {
; CHECK-LABEL: define {{[^@]+}}@kernel0 ; CHECK-LABEL: define {{[^@]+}}@kernel0
; CHECK-SAME: () #[[ATTR0:[0-9]+]] { ; CHECK-SAME: () #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i1 true, i1 false, i1 false) ; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i1 true, i1 false)
; CHECK-NEXT: call void @helper0() ; CHECK-NEXT: call void @helper0()
; CHECK-NEXT: call void @helper1() ; CHECK-NEXT: call void @helper1()
; CHECK-NEXT: call void @helper2() ; CHECK-NEXT: call void @helper2()
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i1 true, i1 false) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i1 true)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
%i = call i32 @__kmpc_target_init(%struct.ident_t* null, i1 true, i1 false, i1 false) %i = call i32 @__kmpc_target_init(%struct.ident_t* null, i1 true, i1 false)
call void @helper0() call void @helper0()
call void @helper1() call void @helper1()
call void @helper2() call void @helper2()
call void @__kmpc_target_deinit(%struct.ident_t* null, i1 true, i1 false) call void @__kmpc_target_deinit(%struct.ident_t* null, i1 true)
ret void ret void
} }
@kernel1_exec_mode = weak constant i8 1 @kernel1_exec_mode = weak constant i8 1
define weak void @kernel1() #0 { define weak void @kernel1() #0 {
; CHECK-LABEL: define {{[^@]+}}@kernel1 ; CHECK-LABEL: define {{[^@]+}}@kernel1
; CHECK-SAME: () #[[ATTR0]] { ; CHECK-SAME: () #[[ATTR0]] {
; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i1 true, i1 false, i1 false) ; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i1 true, i1 false)
; CHECK-NEXT: call void @helper1() ; CHECK-NEXT: call void @helper1()
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i1 false, i1 false) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i1 false)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
%i = call i32 @__kmpc_target_init(%struct.ident_t* null, i1 true, i1 false, i1 false) %i = call i32 @__kmpc_target_init(%struct.ident_t* null, i1 true, i1 false)
call void @helper1() call void @helper1()
call void @__kmpc_target_deinit(%struct.ident_t* null, i1 false, i1 false) call void @__kmpc_target_deinit(%struct.ident_t* null, i1 false)
ret void ret void
} }
@kernel2_exec_mode = weak constant i8 1 @kernel2_exec_mode = weak constant i8 1
define weak void @kernel2() #0 { define weak void @kernel2() #0 {
; CHECK-LABEL: define {{[^@]+}}@kernel2 ; CHECK-LABEL: define {{[^@]+}}@kernel2
; CHECK-SAME: () #[[ATTR0]] { ; CHECK-SAME: () #[[ATTR0]] {
; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i1 false, i1 false, i1 false) ; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i1 false, i1 false)
; CHECK-NEXT: call void @helper0() ; CHECK-NEXT: call void @helper0()
; CHECK-NEXT: call void @helper1() ; CHECK-NEXT: call void @helper1()
; CHECK-NEXT: call void @helper2() ; CHECK-NEXT: call void @helper2()
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i1 false, i1 false) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i1 false)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
%i = call i32 @__kmpc_target_init(%struct.ident_t* null, i1 false, i1 false, i1 false) %i = call i32 @__kmpc_target_init(%struct.ident_t* null, i1 false, i1 false)
call void @helper0() call void @helper0()
call void @helper1() call void @helper1()
call void @helper2() call void @helper2()
call void @__kmpc_target_deinit(%struct.ident_t* null, i1 false, i1 false) call void @__kmpc_target_deinit(%struct.ident_t* null, i1 false)
ret void ret void
} }
define internal void @helper0() { define internal void @helper0() {
; CHECK-LABEL: define {{[^@]+}}@helper0 ; CHECK-LABEL: define {{[^@]+}}@helper0
; CHECK-SAME: () #[[ATTR1:[0-9]+]] { ; CHECK-SAME: () #[[ATTR1:[0-9]+]] {
; CHECK-NEXT: [[THREADLIMIT:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() #[[ATTR1]] ; CHECK-NEXT: [[THREADLIMIT:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() #[[ATTR1]]
; CHECK-NEXT: store i32 [[THREADLIMIT]], i32* @G, align 4 ; CHECK-NEXT: store i32 [[THREADLIMIT]], i32* @G, align 4
Show All 32 Lines
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
%threadLimit = call i32 @__kmpc_get_hardware_num_threads_in_block() %threadLimit = call i32 @__kmpc_get_hardware_num_threads_in_block()
store i32 %threadLimit, i32* @G store i32 %threadLimit, i32* @G
ret void ret void
} }
declare i32 @__kmpc_get_hardware_num_threads_in_block() declare i32 @__kmpc_get_hardware_num_threads_in_block()
declare i32 @__kmpc_target_init(%struct.ident_t*, i1 zeroext, i1 zeroext, i1 zeroext) #1 declare i32 @__kmpc_target_init(%struct.ident_t*, i1 zeroext, i1 zeroext) #1
declare void @__kmpc_target_deinit(%struct.ident_t* nocapture readnone, i1 zeroext, i1 zeroext) #1 declare void @__kmpc_target_deinit(%struct.ident_t* nocapture readnone, i1 zeroext) #1
!llvm.module.flags = !{!0, !1} !llvm.module.flags = !{!0, !1}
!nvvm.annotations = !{!2, !3, !4} !nvvm.annotations = !{!2, !3, !4}
attributes #0 = { optnone noinline "omp_target_thread_limit"="666" "omp_target_num_teams"="777"} attributes #0 = { optnone noinline "omp_target_thread_limit"="666" "omp_target_num_teams"="777"}
!0 = !{i32 7, !"openmp", i32 50} !0 = !{i32 7, !"openmp", i32 50}
Show All 15 Lines

llvm/test/Transforms/OpenMP/global_constructor.ll

; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --check-attributes --include-generated-funcs ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --function-signature --check-attributes --include-generated-funcs
; RUN: opt -S -passes=openmp-opt < %s | FileCheck %s ; RUN: opt -S -passes=openmp-opt < %s | FileCheck %s
%struct.ident_t = type { i32, i32, i32, i32, i8* } %struct.ident_t = type { i32, i32, i32, i32, i8* }
@0 = private unnamed_addr constant [23 x i8] c";unknown;unknown;0;0;;\00", align 1 @0 = private unnamed_addr constant [23 x i8] c";unknown;unknown;0;0;;\00", align 1
@1 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* @0, i32 0, i32 0) }, align 8 @1 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* @0, i32 0, i32 0) }, align 8
@_ZL6Device = internal global double 0.000000e+00, align 8 @_ZL6Device = internal global double 0.000000e+00, align 8
@__omp_offloading_fd02_85283c04_main_l11_exec_mode = weak constant i8 0 @__omp_offloading_fd02_85283c04_main_l11_exec_mode = weak constant i8 0
define weak void @__omp_offloading_fd02_85283c04_main_l11(double* nonnull align 8 dereferenceable(8) %X) local_unnamed_addr { define weak void @__omp_offloading_fd02_85283c04_main_l11(double* nonnull align 8 dereferenceable(8) %X) local_unnamed_addr {
entry: entry:
%0 = tail call i32 @__kmpc_target_init(%struct.ident_t* nonnull @1, i8 2, i1 false, i1 false) #0 %0 = tail call i32 @__kmpc_target_init(%struct.ident_t* nonnull @1, i8 2, i1 false) #0
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %common.ret br i1 %exec_user_code, label %user_code.entry, label %common.ret
common.ret: common.ret:
ret void ret void
user_code.entry: user_code.entry:
%1 = load double, double* @_ZL6Device, align 8, !tbaa !11 %1 = load double, double* @_ZL6Device, align 8, !tbaa !11
%2 = tail call i32 @__kmpc_get_hardware_thread_id_in_block() #0 %2 = tail call i32 @__kmpc_get_hardware_thread_id_in_block() #0
%3 = icmp eq i32 %2, 0 %3 = icmp eq i32 %2, 0
br i1 %3, label %region.guarded, label %region.barrier br i1 %3, label %region.guarded, label %region.barrier
region.guarded: region.guarded:
store double %1, double* %X, align 8, !tbaa !11 store double %1, double* %X, align 8, !tbaa !11
br label %region.barrier br label %region.barrier
region.barrier: region.barrier:
tail call void @__kmpc_barrier_simple_spmd(%struct.ident_t* nonnull @1, i32 %2) tail call void @__kmpc_barrier_simple_spmd(%struct.ident_t* nonnull @1, i32 %2)
tail call void @__kmpc_target_deinit(%struct.ident_t* nonnull @1, i8 2, i1 false) #0 tail call void @__kmpc_target_deinit(%struct.ident_t* nonnull @1, i8 2) #0
br label %common.ret br label %common.ret
} }
declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1, i1) local_unnamed_addr declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1) local_unnamed_addr
declare void @__kmpc_target_deinit(%struct.ident_t*, i8, i1) local_unnamed_addr declare void @__kmpc_target_deinit(%struct.ident_t*, i8) local_unnamed_addr
define internal void @__omp_offloading__fd02_85283c04_Device_l6_ctor() { define internal void @__omp_offloading__fd02_85283c04_Device_l6_ctor() {
entry: entry:
%call.i = tail call double @__nv_log(double noundef 2.000000e+00) #1 %call.i = tail call double @__nv_log(double noundef 2.000000e+00) #1
%call.i2 = tail call double @__nv_log(double noundef 2.000000e+00) #1 %call.i2 = tail call double @__nv_log(double noundef 2.000000e+00) #1
%div = fdiv double %call.i, %call.i2 %div = fdiv double %call.i, %call.i2
store double %div, double* @_ZL6Device, align 8, !tbaa !11 store double %div, double* @_ZL6Device, align 8, !tbaa !11
ret void ret void
Show All 26 Lines
!10 = !{!"clang version 14.0.0"} !10 = !{!"clang version 14.0.0"}
!11 = !{!12, !12, i64 0} !11 = !{!12, !12, i64 0}
!12 = !{!"double", !13, i64 0} !12 = !{!"double", !13, i64 0}
!13 = !{!"omnipotent char", !14, i64 0} !13 = !{!"omnipotent char", !14, i64 0}
!14 = !{!"Simple C++ TBAA"} !14 = !{!"Simple C++ TBAA"}
; CHECK-LABEL: define {{[^@]+}}@__omp_offloading_fd02_85283c04_main_l11 ; CHECK-LABEL: define {{[^@]+}}@__omp_offloading_fd02_85283c04_main_l11
; CHECK-SAME: (double* nonnull align 8 dereferenceable(8) [[X:%.*]]) local_unnamed_addr { ; CHECK-SAME: (double* nonnull align 8 dereferenceable(8) [[X:%.*]]) local_unnamed_addr {
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = tail call i32 @__kmpc_target_init(%struct.ident_t* nonnull @[[GLOB1:[0-9]+]], i8 2, i1 false, i1 false) #[[ATTR1:[0-9]+]] ; CHECK-NEXT: [[TMP0:%.*]] = tail call i32 @__kmpc_target_init(%struct.ident_t* nonnull @[[GLOB1:[0-9]+]], i8 2, i1 false) #[[ATTR2:[0-9]+]]
; CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; CHECK-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; CHECK-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; CHECK: common.ret: ; CHECK: common.ret:
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; CHECK: user_code.entry: ; CHECK: user_code.entry:
; CHECK-NEXT: [[TMP1:%.*]] = load double, double* @_ZL6Device, align 8, !tbaa [[TBAA11:![0-9]+]] ; CHECK-NEXT: [[TMP1:%.*]] = load double, double* @_ZL6Device, align 8, !tbaa [[TBAA11:![0-9]+]]
; CHECK-NEXT: [[TMP2:%.*]] = tail call i32 @__kmpc_get_hardware_thread_id_in_block() #[[ATTR1]] ; CHECK-NEXT: [[TMP2:%.*]] = tail call i32 @__kmpc_get_hardware_thread_id_in_block() #[[ATTR2]]
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i32 [[TMP2]], 0 ; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i32 [[TMP2]], 0
; CHECK-NEXT: br i1 [[TMP3]], label [[REGION_GUARDED:%.*]], label [[REGION_BARRIER:%.*]] ; CHECK-NEXT: br i1 [[TMP3]], label [[REGION_GUARDED:%.*]], label [[REGION_BARRIER:%.*]]
; CHECK: region.guarded: ; CHECK: region.guarded:
; CHECK-NEXT: store double [[TMP1]], double* [[X]], align 8, !tbaa [[TBAA11]] ; CHECK-NEXT: store double [[TMP1]], double* [[X]], align 8, !tbaa [[TBAA11]]
; CHECK-NEXT: br label [[REGION_BARRIER]] ; CHECK-NEXT: br label [[REGION_BARRIER]]
; CHECK: region.barrier: ; CHECK: region.barrier:
; CHECK-NEXT: tail call void @__kmpc_barrier_simple_spmd(%struct.ident_t* nonnull @[[GLOB1]], i32 [[TMP2]]) #[[ATTR1]] ; CHECK-NEXT: tail call void @__kmpc_barrier_simple_spmd(%struct.ident_t* nonnull @[[GLOB1]], i32 [[TMP2]]) #[[ATTR2]]
; CHECK-NEXT: tail call void @__kmpc_target_deinit(%struct.ident_t* nonnull @[[GLOB1]], i8 2, i1 false) #[[ATTR1]] ; CHECK-NEXT: tail call void @__kmpc_target_deinit(%struct.ident_t* nonnull @[[GLOB1]], i8 2) #[[ATTR2]]
; CHECK-NEXT: br label [[COMMON_RET]] ; CHECK-NEXT: br label [[COMMON_RET]]
; ;
;
; CHECK: Function Attrs: norecurse
; CHECK-LABEL: define {{[^@]+}}@__omp_offloading__fd02_85283c04_Device_l6_ctor
; CHECK-SAME: () #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: entry:
; CHECK-NEXT: [[CALL_I:%.*]] = tail call double @__nv_log(double noundef 2.000000e+00) #[[ATTR1:[0-9]+]]
; CHECK-NEXT: [[CALL_I2:%.*]] = tail call double @__nv_log(double noundef 2.000000e+00) #[[ATTR1]]
; CHECK-NEXT: [[DIV:%.*]] = fdiv double [[CALL_I]], [[CALL_I2]]
; CHECK-NEXT: store double [[DIV]], double* @_ZL6Device, align 8, !tbaa [[TBAA11]]
; CHECK-NEXT: ret void
;

llvm/test/Transforms/OpenMP/globalization_remarks.ll

; RUN: opt -passes=openmp-opt -pass-remarks-missed=openmp-opt -disable-output < %s 2>&1 | FileCheck %s ; RUN: opt -passes=openmp-opt -pass-remarks-missed=openmp-opt -disable-output < %s 2>&1 | FileCheck %s
; ModuleID = 'declare_target_codegen_globalization.cpp' ; ModuleID = 'declare_target_codegen_globalization.cpp'
source_filename = "declare_target_codegen_globalization.cpp" source_filename = "declare_target_codegen_globalization.cpp"
target datalayout = "e-i64:64-i128:128-v16:16-v32:32-n16:32:64" target datalayout = "e-i64:64-i128:128-v16:16-v32:32-n16:32:64"
target triple = "nvptx64" target triple = "nvptx64"
; CHECK: remark: globalization_remarks.c:5:7: Could not move globalized variable to the stack. Variable is potentially captured in call. Mark parameter as `__attribute__((noescape))` to override. ; CHECK: remark: globalization_remarks.c:5:7: Could not move globalized variable to the stack. Variable is potentially captured in call. Mark parameter as `__attribute__((noescape))` to override.
; CHECK: remark: globalization_remarks.c:5:7: Found thread data sharing on the GPU. Expect degraded performance due to data globalization. ; CHECK: remark: globalization_remarks.c:5:7: Found thread data sharing on the GPU. Expect degraded performance due to data globalization.
%struct.ident_t = type { i32, i32, i32, i32, i8* } %struct.ident_t = type { i32, i32, i32, i32, i8* }
@S = external local_unnamed_addr global i8* @S = external local_unnamed_addr global i8*
define void @foo() { define void @foo() {
entry: entry:
%c = call i32 @__kmpc_target_init(%struct.ident_t* null, i1 false, i1 true, i1 true) %c = call i32 @__kmpc_target_init(%struct.ident_t* null, i1 false, i1 true)
%0 = call i8* @__kmpc_alloc_shared(i64 4), !dbg !10 %0 = call i8* @__kmpc_alloc_shared(i64 4), !dbg !10
%x_on_stack = bitcast i8* %0 to i32* %x_on_stack = bitcast i8* %0 to i32*
%1 = bitcast i32* %x_on_stack to i8* %1 = bitcast i32* %x_on_stack to i8*
call void @share(i8* %1), !dbg !10 call void @share(i8* %1), !dbg !10
call void @__kmpc_free_shared(i8* %0) call void @__kmpc_free_shared(i8* %0)
call void @__kmpc_target_deinit(%struct.ident_t* null, i1 false, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* null, i1 false)
ret void ret void
} }
define internal void @share(i8* %x) { define internal void @share(i8* %x) {
entry: entry:
store i8* %x, i8** @S store i8* %x, i8** @S
ret void ret void
} }
declare i8* @__kmpc_alloc_shared(i64) declare i8* @__kmpc_alloc_shared(i64)
declare void @__kmpc_free_shared(i8* nocapture) declare void @__kmpc_free_shared(i8* nocapture)
declare i32 @__kmpc_target_init(%struct.ident_t*, i1, i1, i1); declare i32 @__kmpc_target_init(%struct.ident_t*, i1, i1);
declare void @__kmpc_target_deinit(%struct.ident_t*, i1, i1) declare void @__kmpc_target_deinit(%struct.ident_t*, i1)
!llvm.dbg.cu = !{!0} !llvm.dbg.cu = !{!0}
!llvm.module.flags = !{!3, !4, !5, !6} !llvm.module.flags = !{!3, !4, !5, !6}
!nvvm.annotations = !{!7, !8} !nvvm.annotations = !{!7, !8}
!0 = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 12.0.0", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, enums: !2, splitDebugInlining: false, nameTableKind: None) !0 = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 12.0.0", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, enums: !2, splitDebugInlining: false, nameTableKind: None)
!1 = !DIFile(filename: "globalization_remarks.c", directory: "/tmp/globalization_remarks.c") !1 = !DIFile(filename: "globalization_remarks.c", directory: "/tmp/globalization_remarks.c")
!2 = !{} !2 = !{}
!3 = !{i32 2, !"Debug Info Version", i32 3} !3 = !{i32 2, !"Debug Info Version", i32 3}
!4 = !{i32 1, !"wchar_size", i32 4} !4 = !{i32 1, !"wchar_size", i32 4}
!5 = !{i32 7, !"openmp", i32 50} !5 = !{i32 7, !"openmp", i32 50}
!6 = !{i32 7, !"openmp-device", i32 50} !6 = !{i32 7, !"openmp-device", i32 50}
!7 = !{void ()* @foo, !"kernel", i32 1} !7 = !{void ()* @foo, !"kernel", i32 1}
!8 = distinct !DISubprogram(name: "foo", scope: !1, file: !1, line: 1, type: !9, scopeLine: 1, flags: DIFlagPrototyped, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !0, retainedNodes: !2) !8 = distinct !DISubprogram(name: "foo", scope: !1, file: !1, line: 1, type: !9, scopeLine: 1, flags: DIFlagPrototyped, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !0, retainedNodes: !2)
!9 = !DISubroutineType(types: !2) !9 = !DISubroutineType(types: !2)
!10 = !DILocation(line: 5, column: 7, scope: !8) !10 = !DILocation(line: 5, column: 7, scope: !8)

llvm/test/Transforms/OpenMP/gpu_state_machine_function_ptr_replacement.ll

Show First 20 Lines • Show All 42 Lines • ▼ Show 20 Lines
@2 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 2, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* @0, i32 0, i32 0) }, align 8 @2 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 2, i32 0, i8* getelementptr inbounds ([23 x i8], [23 x i8]* @0, i32 0, i32 0) }, align 8
@llvm.compiler.used = appending global [1 x i8*] [i8* @__omp_offloading_10301_87b2c_foo_l7_exec_mode], section "llvm.metadata" @llvm.compiler.used = appending global [1 x i8*] [i8* @__omp_offloading_10301_87b2c_foo_l7_exec_mode], section "llvm.metadata"
define weak void @__omp_offloading_10301_87b2c_foo_l7() { define weak void @__omp_offloading_10301_87b2c_foo_l7() {
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 %1, i32* %.threadid_temp., align 4 store i32 %1, i32* %.threadid_temp., align 4
call void @__omp_outlined__(i32* %.threadid_temp., i32* %.zero.addr) call void @__omp_outlined__(i32* %.threadid_temp., i32* %.zero.addr)
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1, i1) { define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1) {
ret i32 0 ret i32 0
} }
declare void @unknown() declare void @unknown()
define internal void @__omp_outlined__(i32* noalias %.global_tid., i32* noalias %.bound_tid.) { define internal void @__omp_outlined__(i32* noalias %.global_tid., i32* noalias %.bound_tid.) {
entry: entry:
%.global_tid..addr = alloca i32*, align 8 %.global_tid..addr = alloca i32*, align 8
▲ Show 20 Lines • Show All 69 Lines • ▼ Show 20 Linesentry:
store i32 %1, i32* %.addr1, align 4 store i32 %1, i32* %.addr1, align 4
call void @__kmpc_get_shared_variables(i8*** %global_args) call void @__kmpc_get_shared_variables(i8*** %global_args)
call void @__omp_outlined__2(i32* %.addr1, i32* %.zero.addr) call void @__omp_outlined__2(i32* %.addr1, i32* %.zero.addr)
ret void ret void
} }
declare i32 @__kmpc_global_thread_num(%struct.ident_t*) declare i32 @__kmpc_global_thread_num(%struct.ident_t*)
declare void @__kmpc_target_deinit(%struct.ident_t*, i8, i1) declare void @__kmpc_target_deinit(%struct.ident_t*, i8)
define internal void @__omp_outlined__3(i32* noalias %.global_tid., i32* noalias %.bound_tid.) { define internal void @__omp_outlined__3(i32* noalias %.global_tid., i32* noalias %.bound_tid.) {
entry: entry:
%.global_tid..addr = alloca i32*, align 8 %.global_tid..addr = alloca i32*, align 8
%.bound_tid..addr = alloca i32*, align 8 %.bound_tid..addr = alloca i32*, align 8
store i32* %.global_tid., i32** %.global_tid..addr, align 8 store i32* %.global_tid., i32** %.global_tid..addr, align 8
store i32* %.bound_tid., i32** %.bound_tid..addr, align 8 store i32* %.bound_tid., i32** %.bound_tid..addr, align 8
ret void ret void
Show All 24 Lines

llvm/test/Transforms/OpenMP/is_spmd_exec_mode_fold.ll

Show All 15 Lines
; CHECK: @[[WILL_BE_SPMD_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 3 ; CHECK: @[[WILL_BE_SPMD_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 3
; CHECK: @[[NON_SPMD_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 1 ; CHECK: @[[NON_SPMD_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 1
; CHECK: @[[WILL_NOT_BE_SPMD_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 1 ; CHECK: @[[WILL_NOT_BE_SPMD_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 1
; CHECK: @[[G:[a-zA-Z0-9_$"\\.-]+]] = external global i8 ; CHECK: @[[G:[a-zA-Z0-9_$"\\.-]+]] = external global i8
; CHECK: @[[LLVM_COMPILER_USED:[a-zA-Z0-9_$"\\.-]+]] = appending global [4 x i8*] [i8* @is_spmd_exec_mode, i8* @will_be_spmd_exec_mode, i8* @non_spmd_exec_mode, i8* @will_not_be_spmd_exec_mode], section "llvm.metadata" ; CHECK: @[[LLVM_COMPILER_USED:[a-zA-Z0-9_$"\\.-]+]] = appending global [4 x i8*] [i8* @is_spmd_exec_mode, i8* @will_be_spmd_exec_mode, i8* @non_spmd_exec_mode, i8* @will_not_be_spmd_exec_mode], section "llvm.metadata"
;. ;.
define weak void @is_spmd() { define weak void @is_spmd() {
; CHECK-LABEL: define {{[^@]+}}@is_spmd() { ; CHECK-LABEL: define {{[^@]+}}@is_spmd() {
; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false, i1 false) ; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false)
; CHECK-NEXT: call void @is_spmd_helper1() ; CHECK-NEXT: call void @is_spmd_helper1()
; CHECK-NEXT: call void @is_spmd_helper2() ; CHECK-NEXT: call void @is_spmd_helper2()
; CHECK-NEXT: call void @is_mixed_helper() ; CHECK-NEXT: call void @is_mixed_helper()
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2, i1 false) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
%i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false, i1 false) %i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false)
call void @is_spmd_helper1() call void @is_spmd_helper1()
call void @is_spmd_helper2() call void @is_spmd_helper2()
call void @is_mixed_helper() call void @is_mixed_helper()
call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2, i1 false) call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2)
ret void ret void
} }
define weak void @will_be_spmd() { define weak void @will_be_spmd() {
; CHECK-LABEL: define {{[^@]+}}@will_be_spmd() { ; CHECK-LABEL: define {{[^@]+}}@will_be_spmd() {
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x i8*], align 8 ; CHECK-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x i8*], align 8
; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false, i1 false) ; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false)
; CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[I]], -1 ; CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[I]], -1
; CHECK-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; CHECK-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; CHECK: common.ret: ; CHECK: common.ret:
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; CHECK: user_code.entry: ; CHECK: user_code.entry:
; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* null) #[[ATTR2:[0-9]+]] ; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* null) #[[ATTR2:[0-9]+]]
; CHECK-NEXT: [[TMP1:%.*]] = bitcast [0 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8** ; CHECK-NEXT: [[TMP1:%.*]] = bitcast [0 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8**
; CHECK-NEXT: call void @is_spmd_helper2() ; CHECK-NEXT: call void @is_spmd_helper2()
; CHECK-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* null, i32 [[TMP0]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__ to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined___wrapper to i8*), i8** [[TMP1]], i64 0) ; CHECK-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* null, i32 [[TMP0]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__ to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined___wrapper to i8*), i8** [[TMP1]], i64 0)
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2, i1 false) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
entry: entry:
%captured_vars_addrs = alloca [0 x i8*], align 8 %captured_vars_addrs = alloca [0 x i8*], align 8
%i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 true, i1 true) %i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 true)
%exec_user_code = icmp eq i32 %i, -1 %exec_user_code = icmp eq i32 %i, -1
br i1 %exec_user_code, label %user_code.entry, label %common.ret br i1 %exec_user_code, label %user_code.entry, label %common.ret
common.ret: common.ret:
ret void ret void
user_code.entry: user_code.entry:
%0 = call i32 @__kmpc_global_thread_num(%struct.ident_t* null) %0 = call i32 @__kmpc_global_thread_num(%struct.ident_t* null)
%1 = bitcast [0 x i8*]* %captured_vars_addrs to i8** %1 = bitcast [0 x i8*]* %captured_vars_addrs to i8**
call void @is_spmd_helper2() call void @is_spmd_helper2()
call void @__kmpc_parallel_51(%struct.ident_t* null, i32 %0, i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__ to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined___wrapper to i8*), i8** %1, i64 0) call void @__kmpc_parallel_51(%struct.ident_t* null, i32 %0, i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__ to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined___wrapper to i8*), i8** %1, i64 0)
call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1)
ret void ret void
} }
define weak void @non_spmd() { define weak void @non_spmd() {
; CHECK-LABEL: define {{[^@]+}}@non_spmd() { ; CHECK-LABEL: define {{[^@]+}}@non_spmd() {
; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 false, i1 false) ; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 false)
; CHECK-NEXT: call void @is_generic_helper1() ; CHECK-NEXT: call void @is_generic_helper1()
; CHECK-NEXT: call void @is_generic_helper2() ; CHECK-NEXT: call void @is_generic_helper2()
; CHECK-NEXT: call void @is_mixed_helper() ; CHECK-NEXT: call void @is_mixed_helper()
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1, i1 false) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
%i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 false, i1 false) %i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 false)
call void @is_generic_helper1() call void @is_generic_helper1()
call void @is_generic_helper2() call void @is_generic_helper2()
call void @is_mixed_helper() call void @is_mixed_helper()
call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1, i1 false) call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1)
ret void ret void
} }
define weak void @will_not_be_spmd() { define weak void @will_not_be_spmd() {
; CHECK-LABEL: define {{[^@]+}}@will_not_be_spmd() { ; CHECK-LABEL: define {{[^@]+}}@will_not_be_spmd() {
; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 false, i1 false) ; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 false)
; CHECK-NEXT: call void @is_generic_helper1() ; CHECK-NEXT: call void @is_generic_helper1()
; CHECK-NEXT: call void @is_generic_helper2() ; CHECK-NEXT: call void @is_generic_helper2()
; CHECK-NEXT: call void @is_mixed_helper() ; CHECK-NEXT: call void @is_mixed_helper()
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1, i1 false) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
%i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 false, i1 false) %i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 false)
call void @is_generic_helper1() call void @is_generic_helper1()
call void @is_generic_helper2() call void @is_generic_helper2()
call void @is_mixed_helper() call void @is_mixed_helper()
call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1, i1 false) call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1)
ret void ret void
} }
define internal void @is_spmd_helper1() { define internal void @is_spmd_helper1() {
; CHECK-LABEL: define {{[^@]+}}@is_spmd_helper1() { ; CHECK-LABEL: define {{[^@]+}}@is_spmd_helper1() {
; CHECK-NEXT: store i8 1, i8* @G, align 1 ; CHECK-NEXT: store i8 1, i8* @G, align 1
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
▲ Show 20 Lines • Show All 80 Lines • ▼ Show 20 Lines
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
entry: entry:
ret void ret void
} }
declare void @spmd_compatible() "llvm.assume"="ompx_spmd_amenable" declare void @spmd_compatible() "llvm.assume"="ompx_spmd_amenable"
declare i8 @__kmpc_is_spmd_exec_mode() declare i8 @__kmpc_is_spmd_exec_mode()
declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1 zeroext, i1 zeroext) declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1 zeroext)
declare void @__kmpc_target_deinit(%struct.ident_t* nocapture readnone, i8, i1 zeroext) declare void @__kmpc_target_deinit(%struct.ident_t* nocapture readnone, i8)
declare void @__kmpc_parallel_51(%struct.ident_t*, i32, i32, i32, i32, i8*, i8*, i8**, i64) declare void @__kmpc_parallel_51(%struct.ident_t*, i32, i32, i32, i32, i8*, i8*, i8**, i64)
declare i32 @__kmpc_global_thread_num(%struct.ident_t*) declare i32 @__kmpc_global_thread_num(%struct.ident_t*)
declare void @foo() declare void @foo()
declare void @bar() declare void @bar()
!llvm.module.flags = !{!0, !1} !llvm.module.flags = !{!0, !1}
!nvvm.annotations = !{!2, !3, !4, !5} !nvvm.annotations = !{!2, !3, !4, !5}
Show All 18 Lines

llvm/test/Transforms/OpenMP/parallel_level_fold.ll

Show All 13 Lines
; CHECK: @[[NO_SPMD_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 1 ; CHECK: @[[NO_SPMD_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 1
; CHECK: @[[SPMD_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 0 ; CHECK: @[[SPMD_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 0
; CHECK: @[[PARALLEL_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 0 ; CHECK: @[[PARALLEL_EXEC_MODE:[a-zA-Z0-9_$"\\.-]+]] = weak constant i8 0
; CHECK: @[[G:[a-zA-Z0-9_$"\\.-]+]] = external global i8 ; CHECK: @[[G:[a-zA-Z0-9_$"\\.-]+]] = external global i8
; CHECK: @[[LLVM_COMPILER_USED:[a-zA-Z0-9_$"\\.-]+]] = appending global [3 x i8*] [i8* @no_spmd_exec_mode, i8* @spmd_exec_mode, i8* @parallel_exec_mode], section "llvm.metadata" ; CHECK: @[[LLVM_COMPILER_USED:[a-zA-Z0-9_$"\\.-]+]] = appending global [3 x i8*] [i8* @no_spmd_exec_mode, i8* @spmd_exec_mode, i8* @parallel_exec_mode], section "llvm.metadata"
;. ;.
define weak void @none_spmd() { define weak void @none_spmd() {
; CHECK-LABEL: define {{[^@]+}}@none_spmd() { ; CHECK-LABEL: define {{[^@]+}}@none_spmd() {
; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 false, i1 false) ; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 false)
; CHECK-NEXT: call void @none_spmd_helper() ; CHECK-NEXT: call void @none_spmd_helper()
; CHECK-NEXT: call void @mixed_helper() ; CHECK-NEXT: call void @mixed_helper()
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1, i1 false) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
%i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 false, i1 false) %i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 1, i1 false)
call void @none_spmd_helper() call void @none_spmd_helper()
call void @mixed_helper() call void @mixed_helper()
call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1, i1 false) call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1)
ret void ret void
} }
define weak void @spmd() { define weak void @spmd() {
; CHECK-LABEL: define {{[^@]+}}@spmd() { ; CHECK-LABEL: define {{[^@]+}}@spmd() {
; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false, i1 false) ; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false)
; CHECK-NEXT: call void @spmd_helper() ; CHECK-NEXT: call void @spmd_helper()
; CHECK-NEXT: call void @mixed_helper() ; CHECK-NEXT: call void @mixed_helper()
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2, i1 false) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
%i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false, i1 false) %i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false)
call void @spmd_helper() call void @spmd_helper()
call void @mixed_helper() call void @mixed_helper()
call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2, i1 false) call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2)
ret void ret void
} }
define weak void @parallel() { define weak void @parallel() {
; CHECK-LABEL: define {{[^@]+}}@parallel() { ; CHECK-LABEL: define {{[^@]+}}@parallel() {
; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false, i1 false) ; CHECK-NEXT: [[I:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false)
; CHECK-NEXT: call void @spmd_helper() ; CHECK-NEXT: call void @spmd_helper()
; CHECK-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* null, i32 0, i32 0, i32 0, i32 0, i8* null, i8* null, i8** null, i64 0) ; CHECK-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* null, i32 0, i32 0, i32 0, i32 0, i8* null, i8* null, i8** null, i64 0)
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2, i1 false) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
%i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false, i1 false) %i = call i32 @__kmpc_target_init(%struct.ident_t* null, i8 2, i1 false)
call void @spmd_helper() call void @spmd_helper()
call void @__kmpc_parallel_51(%struct.ident_t* null, i32 0, i32 0, i32 0, i32 0, i8* null, i8* null, i8** null, i64 0) call void @__kmpc_parallel_51(%struct.ident_t* null, i32 0, i32 0, i32 0, i32 0, i8* null, i8* null, i8** null, i64 0)
call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2, i1 false) call void @__kmpc_target_deinit(%struct.ident_t* null, i8 2)
ret void ret void
} }
define internal void @mixed_helper() { define internal void @mixed_helper() {
; CHECK-LABEL: define {{[^@]+}}@mixed_helper() { ; CHECK-LABEL: define {{[^@]+}}@mixed_helper() {
; CHECK-NEXT: [[LEVEL:%.*]] = call i8 @__kmpc_parallel_level() ; CHECK-NEXT: [[LEVEL:%.*]] = call i8 @__kmpc_parallel_level()
; CHECK-NEXT: store i8 [[LEVEL]], i8* @G, align 1 ; CHECK-NEXT: store i8 [[LEVEL]], i8* @G, align 1
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
▲ Show 20 Lines • Show All 55 Lines • ▼ Show 20 Lines;
%level = call i8 @__kmpc_parallel_level() %level = call i8 @__kmpc_parallel_level()
store i8 %level, i8* @G store i8 %level, i8* @G
ret void ret void
} }
declare void @foo() declare void @foo()
declare void @bar() declare void @bar()
declare i8 @__kmpc_parallel_level() declare i8 @__kmpc_parallel_level()
declare i32 @__kmpc_target_init(%struct.ident_t*, i8 zeroext, i1 zeroext, i1 zeroext) #1 declare i32 @__kmpc_target_init(%struct.ident_t*, i8 zeroext, i1 zeroext) #1
declare void @__kmpc_target_deinit(%struct.ident_t* nocapture readnone, i8 zeroext, i1 zeroext) #1 declare void @__kmpc_target_deinit(%struct.ident_t* nocapture readnone, i8 zeroext) #1
!llvm.module.flags = !{!0, !1} !llvm.module.flags = !{!0, !1}
!nvvm.annotations = !{!2, !3, !4} !nvvm.annotations = !{!2, !3, !4}
!0 = !{i32 7, !"openmp", i32 50} !0 = !{i32 7, !"openmp", i32 50}
!1 = !{i32 7, !"openmp-device", i32 50} !1 = !{i32 7, !"openmp-device", i32 50}
!2 = !{void ()* @none_spmd, !"kernel", i32 1} !2 = !{void ()* @none_spmd, !"kernel", i32 1}
!3 = !{void ()* @spmd, !"kernel", i32 1} !3 = !{void ()* @spmd, !"kernel", i32 1}
Show All 10 Lines

llvm/test/Transforms/OpenMP/remove_globalization.ll

Show All 11 Lines
; CHECK-REMARKS: remark: remove_globalization.c:4:2: Found thread data sharing on the GPU. Expect degraded performance due to data globalization. ; CHECK-REMARKS: remark: remove_globalization.c:4:2: Found thread data sharing on the GPU. Expect degraded performance due to data globalization.
; UTC_ARGS: --enable ; UTC_ARGS: --enable
@S = external local_unnamed_addr global i8* @S = external local_unnamed_addr global i8*
%struct.ident_t = type { i32, i32, i32, i32, i8* } %struct.ident_t = type { i32, i32, i32, i32, i8* }
; Make it a weak definition so we will apply custom state machine rewriting but can't use the body in the reasoning. ; Make it a weak definition so we will apply custom state machine rewriting but can't use the body in the reasoning.
define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1, i1) {
ret i32 0
}
declare void @__kmpc_target_deinit(%struct.ident_t*, i8, i1)
;. ;.
; CHECK: @[[S:[a-zA-Z0-9_$"\\.-]+]] = external local_unnamed_addr global i8* ; CHECK: @[[S:[a-zA-Z0-9_$"\\.-]+]] = external local_unnamed_addr global i8*
;. ;.
; CHECK-DISABLED: @[[S:[a-zA-Z0-9_$"\\.-]+]] = external local_unnamed_addr global i8* ; CHECK-DISABLED: @[[S:[a-zA-Z0-9_$"\\.-]+]] = external local_unnamed_addr global i8*
;. ;.
define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1) {
; CHECK-LABEL: define {{[^@]+}}@__kmpc_target_init
; CHECK-SAME: (%struct.ident_t* [[TMP0:%.*]], i8 [[TMP1:%.*]], i1 [[TMP2:%.*]]) {
; CHECK-NEXT: ret i32 0
;
; CHECK-DISABLED-LABEL: define {{[^@]+}}@__kmpc_target_init
; CHECK-DISABLED-SAME: (%struct.ident_t* [[TMP0:%.*]], i8 [[TMP1:%.*]], i1 [[TMP2:%.*]]) {
; CHECK-DISABLED-NEXT: ret i32 0
;
ret i32 0
}
declare void @__kmpc_target_deinit(%struct.ident_t*, i8)
define void @kernel() { define void @kernel() {
; CHECK-LABEL: define {{[^@]+}}@kernel() { ; CHECK-LABEL: define {{[^@]+}}@kernel() {
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* nonnull null, i8 1, i1 false, i1 true) ; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* nonnull null, i8 1, i1 false)
; CHECK-NEXT: call void @foo() #[[ATTR0:[0-9]+]] ; CHECK-NEXT: call void @foo() #[[ATTR0:[0-9]+]]
; CHECK-NEXT: call void @bar() #[[ATTR0]] ; CHECK-NEXT: call void @bar() #[[ATTR0]]
; CHECK-NEXT: call void @convert_and_move_alloca() #[[ATTR0]] ; CHECK-NEXT: call void @convert_and_move_alloca() #[[ATTR0]]
; CHECK-NEXT: call void @unknown_no_openmp() ; CHECK-NEXT: call void @unknown_no_openmp()
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* nonnull null, i8 1, i1 true) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* nonnull null, i8 1)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
; CHECK-DISABLED-LABEL: define {{[^@]+}}@kernel() { ; CHECK-DISABLED-LABEL: define {{[^@]+}}@kernel() {
; CHECK-DISABLED-NEXT: entry: ; CHECK-DISABLED-NEXT: entry:
; CHECK-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* nonnull null, i8 1, i1 false, i1 true) ; CHECK-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* nonnull null, i8 1, i1 false)
; CHECK-DISABLED-NEXT: call void @foo() #[[ATTR0:[0-9]+]] ; CHECK-DISABLED-NEXT: call void @foo() #[[ATTR0:[0-9]+]]
; CHECK-DISABLED-NEXT: call void @bar() #[[ATTR0]] ; CHECK-DISABLED-NEXT: call void @bar() #[[ATTR0]]
; CHECK-DISABLED-NEXT: call void @convert_and_move_alloca() #[[ATTR0]] ; CHECK-DISABLED-NEXT: call void @convert_and_move_alloca() #[[ATTR0]]
; CHECK-DISABLED-NEXT: call void @unknown_no_openmp() ; CHECK-DISABLED-NEXT: call void @unknown_no_openmp()
; CHECK-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* nonnull null, i8 1, i1 true) ; CHECK-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* nonnull null, i8 1)
; CHECK-DISABLED-NEXT: ret void ; CHECK-DISABLED-NEXT: ret void
; ;
entry: entry:
%0 = call i32 @__kmpc_target_init(%struct.ident_t* nonnull null, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* nonnull null, i8 1, i1 true)
call void @foo() call void @foo()
call void @bar() call void @bar()
call void @convert_and_move_alloca() call void @convert_and_move_alloca()
call void @unknown_no_openmp() call void @unknown_no_openmp()
call void @__kmpc_target_deinit(%struct.ident_t* nonnull null, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* nonnull null, i8 1)
ret void ret void
} }
define internal void @foo() { define internal void @foo() {
; CHECK-LABEL: define {{[^@]+}}@foo ; CHECK-LABEL: define {{[^@]+}}@foo
; CHECK-SAME: () #[[ATTR0]] { ; CHECK-SAME: () #[[ATTR0]] {
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: [[DOTH2S:%.*]] = alloca i8, i64 4, align 1 ; CHECK-NEXT: [[DOTH2S:%.*]] = alloca i8, i64 4, align 1
▲ Show 20 Lines • Show All 223 Lines • ▼ Show 20 Lines
; CHECK-DISABLED: [[META5:![0-9]+]] = !{i32 7, !"openmp", i32 50} ; CHECK-DISABLED: [[META5:![0-9]+]] = !{i32 7, !"openmp", i32 50}
; CHECK-DISABLED: [[META6:![0-9]+]] = !{i32 7, !"openmp-device", i32 50} ; CHECK-DISABLED: [[META6:![0-9]+]] = !{i32 7, !"openmp-device", i32 50}
; CHECK-DISABLED: [[META7:![0-9]+]] = !{void ()* @kernel, !"kernel", i32 1} ; CHECK-DISABLED: [[META7:![0-9]+]] = !{void ()* @kernel, !"kernel", i32 1}
; CHECK-DISABLED: [[DBG8]] = !DILocation(line: 4, column: 2, scope: !9) ; CHECK-DISABLED: [[DBG8]] = !DILocation(line: 4, column: 2, scope: !9)
; CHECK-DISABLED: [[META9:![0-9]+]] = distinct !DISubprogram(name: "bar", scope: !1, file: !1, line: 1, type: !10, scopeLine: 1, flags: DIFlagPrototyped, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !0, retainedNodes: !2) ; CHECK-DISABLED: [[META9:![0-9]+]] = distinct !DISubprogram(name: "bar", scope: !1, file: !1, line: 1, type: !10, scopeLine: 1, flags: DIFlagPrototyped, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !0, retainedNodes: !2)
; CHECK-DISABLED: [[META10:![0-9]+]] = !DISubroutineType(types: !2) ; CHECK-DISABLED: [[META10:![0-9]+]] = !DISubroutineType(types: !2)
; CHECK-DISABLED: [[DBG11]] = !DILocation(line: 6, column: 2, scope: !9) ; CHECK-DISABLED: [[DBG11]] = !DILocation(line: 6, column: 2, scope: !9)
;. ;.
;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
; CHECK-REMARKS: {{.*}}

llvm/test/Transforms/OpenMP/replace_globalization.ll

Show All 19 Lines
@1 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([113 x i8], [113 x i8]* @0, i32 0, i32 0) }, align 8 @1 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 0, i32 0, i8* getelementptr inbounds ([113 x i8], [113 x i8]* @0, i32 0, i32 0) }, align 8
@foo_exec_mode = weak constant i8 1 @foo_exec_mode = weak constant i8 1
@bar_exec_mode = weak constant i8 1 @bar_exec_mode = weak constant i8 1
@baz_spmd_exec_mode = weak constant i8 2 @baz_spmd_exec_mode = weak constant i8 2
define dso_local void @foo() "kernel" { define dso_local void @foo() "kernel" {
entry: entry:
%c = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %c = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%x = call align 4 i8* @__kmpc_alloc_shared(i64 4) %x = call align 4 i8* @__kmpc_alloc_shared(i64 4)
call void @unknown_no_openmp() call void @unknown_no_openmp()
%x_on_stack = bitcast i8* %x to i32* %x_on_stack = bitcast i8* %x to i32*
%0 = bitcast i32* %x_on_stack to i8* %0 = bitcast i32* %x_on_stack to i8*
call void @use(i8* %0) call void @use(i8* %0)
call void @__kmpc_free_shared(i8* %x, i64 4) call void @__kmpc_free_shared(i8* %x, i64 4)
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
} }
define void @bar() "kernel" { define void @bar() "kernel" {
%c = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %c = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
call void @unknown_no_openmp() call void @unknown_no_openmp()
%cmp = icmp eq i32 %c, -1 %cmp = icmp eq i32 %c, -1
br i1 %cmp, label %master1, label %exit br i1 %cmp, label %master1, label %exit
master1: master1:
%x = call align 4 i8* @__kmpc_alloc_shared(i64 16), !dbg !11 %x = call align 4 i8* @__kmpc_alloc_shared(i64 16), !dbg !11
%x_on_stack = bitcast i8* %x to [4 x i32]* %x_on_stack = bitcast i8* %x to [4 x i32]*
%a0 = bitcast [4 x i32]* %x_on_stack to i8* %a0 = bitcast [4 x i32]* %x_on_stack to i8*
call void @use(i8* %a0) call void @use(i8* %a0)
call void @__kmpc_free_shared(i8* %x, i64 16) call void @__kmpc_free_shared(i8* %x, i64 16)
br label %next br label %next
next: next:
call void @unknown_no_openmp() call void @unknown_no_openmp()
%b0 = icmp eq i32 %c, -1 %b0 = icmp eq i32 %c, -1
br i1 %b0, label %master2, label %exit br i1 %b0, label %master2, label %exit
master2: master2:
%y = call align 4 i8* @__kmpc_alloc_shared(i64 4), !dbg !12 %y = call align 4 i8* @__kmpc_alloc_shared(i64 4), !dbg !12
%y_on_stack = bitcast i8* %y to [4 x i32]* %y_on_stack = bitcast i8* %y to [4 x i32]*
%b1 = bitcast [4 x i32]* %y_on_stack to i8* %b1 = bitcast [4 x i32]* %y_on_stack to i8*
call void @use(i8* %b1) call void @use(i8* %b1)
call void @__kmpc_free_shared(i8* %y, i64 4) call void @__kmpc_free_shared(i8* %y, i64 4)
br label %exit br label %exit
exit: exit:
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
} }
define void @baz_spmd() "kernel" { define void @baz_spmd() "kernel" {
%c = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 2, i1 true, i1 true) %c = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 2, i1 true)
call void @unknown_no_openmp() call void @unknown_no_openmp()
%c0 = icmp eq i32 %c, -1 %c0 = icmp eq i32 %c, -1
br i1 %c0, label %master3, label %exit br i1 %c0, label %master3, label %exit
master3: master3:
%z = call align 4 i8* @__kmpc_alloc_shared(i64 24), !dbg !12 %z = call align 4 i8* @__kmpc_alloc_shared(i64 24), !dbg !12
%z_on_stack = bitcast i8* %z to [6 x i32]* %z_on_stack = bitcast i8* %z to [6 x i32]*
%c1 = bitcast [6 x i32]* %z_on_stack to i8* %c1 = bitcast [6 x i32]* %z_on_stack to i8*
call void @use(i8* %c1) call void @use(i8* %c1)
call void @__kmpc_free_shared(i8* %z, i64 24) call void @__kmpc_free_shared(i8* %z, i64 24)
br label %exit br label %exit
exit: exit:
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 2, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 2)
ret void ret void
} }
define void @use(i8* %x) { define void @use(i8* %x) {
entry: entry:
store i8* %x, i8** @S store i8* %x, i8** @S
ret void ret void
} }
Show All 12 Lines
declare i32 @llvm.nvvm.read.ptx.sreg.tid.x() declare i32 @llvm.nvvm.read.ptx.sreg.tid.x()
declare i32 @llvm.nvvm.read.ptx.sreg.ntid.x() declare i32 @llvm.nvvm.read.ptx.sreg.ntid.x()
declare i32 @llvm.nvvm.read.ptx.sreg.warpsize() declare i32 @llvm.nvvm.read.ptx.sreg.warpsize()
; Make it a weak definition so we will apply custom state machine rewriting but can't use the body in the reasoning. ; Make it a weak definition so we will apply custom state machine rewriting but can't use the body in the reasoning.
define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1, i1) { define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1) {
ret i32 0 ret i32 0
} }
declare void @__kmpc_target_deinit(%struct.ident_t*, i8, i1) declare void @__kmpc_target_deinit(%struct.ident_t*, i8)
declare void @unknown_no_openmp() "llvm.assume"="omp_no_openmp" declare void @unknown_no_openmp() "llvm.assume"="omp_no_openmp"
!llvm.dbg.cu = !{!0} !llvm.dbg.cu = !{!0}
!llvm.module.flags = !{!3, !4, !5, !6} !llvm.module.flags = !{!3, !4, !5, !6}
!nvvm.annotations = !{!7, !8, !13} !nvvm.annotations = !{!7, !8, !13}
!0 = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 12.0.0", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, enums: !2, splitDebugInlining: false, nameTableKind: None) !0 = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 12.0.0", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, enums: !2, splitDebugInlining: false, nameTableKind: None)
Show All 20 Lines
; CHECK: @[[OFFSET:[a-zA-Z0-9_$"\\.-]+]] = global i32 undef ; CHECK: @[[OFFSET:[a-zA-Z0-9_$"\\.-]+]] = global i32 undef
; CHECK: @[[STACK:[a-zA-Z0-9_$"\\.-]+]] = internal addrspace(3) global [1024 x i8] undef ; CHECK: @[[STACK:[a-zA-Z0-9_$"\\.-]+]] = internal addrspace(3) global [1024 x i8] undef
; CHECK: @[[X_SHARED:[a-zA-Z0-9_$"\\.-]+]] = internal addrspace(3) global [16 x i8] undef, align 4 ; CHECK: @[[X_SHARED:[a-zA-Z0-9_$"\\.-]+]] = internal addrspace(3) global [16 x i8] undef, align 4
; CHECK: @[[Y_SHARED:[a-zA-Z0-9_$"\\.-]+]] = internal addrspace(3) global [4 x i8] undef, align 4 ; CHECK: @[[Y_SHARED:[a-zA-Z0-9_$"\\.-]+]] = internal addrspace(3) global [4 x i8] undef, align 4
;. ;.
; CHECK-LABEL: define {{[^@]+}}@foo ; CHECK-LABEL: define {{[^@]+}}@foo
; CHECK-SAME: () #[[ATTR0:[0-9]+]] { ; CHECK-SAME: () #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: [[C:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; CHECK-NEXT: [[C:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; CHECK-NEXT: [[X:%.*]] = call align 4 i8* @__kmpc_alloc_shared(i64 4) #[[ATTR6:[0-9]+]] ; CHECK-NEXT: [[X:%.*]] = call align 4 i8* @__kmpc_alloc_shared(i64 4) #[[ATTR6:[0-9]+]]
; CHECK-NEXT: call void @unknown_no_openmp() ; CHECK-NEXT: call void @unknown_no_openmp()
; CHECK-NEXT: call void @use.internalized(i8* nofree [[X]]) #[[ATTR6]] ; CHECK-NEXT: call void @use.internalized(i8* nofree [[X]]) #[[ATTR6]]
; CHECK-NEXT: call void @__kmpc_free_shared(i8* [[X]], i64 4) #[[ATTR6]] ; CHECK-NEXT: call void @__kmpc_free_shared(i8* [[X]], i64 4) #[[ATTR6]]
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
; ;
; CHECK-LABEL: define {{[^@]+}}@bar ; CHECK-LABEL: define {{[^@]+}}@bar
; CHECK-SAME: () #[[ATTR0]] { ; CHECK-SAME: () #[[ATTR0]] {
; CHECK-NEXT: [[C:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; CHECK-NEXT: [[C:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; CHECK-NEXT: call void @unknown_no_openmp() ; CHECK-NEXT: call void @unknown_no_openmp()
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[C]], -1 ; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[C]], -1
; CHECK-NEXT: br i1 [[CMP]], label [[MASTER1:%.*]], label [[EXIT:%.*]] ; CHECK-NEXT: br i1 [[CMP]], label [[MASTER1:%.*]], label [[EXIT:%.*]]
; CHECK: master1: ; CHECK: master1:
; CHECK-NEXT: call void @use.internalized(i8* nofree addrspacecast (i8 addrspace(3)* getelementptr inbounds ([16 x i8], [16 x i8] addrspace(3)* @x_shared, i32 0, i32 0) to i8*)) #[[ATTR6]] ; CHECK-NEXT: call void @use.internalized(i8* nofree addrspacecast (i8 addrspace(3)* getelementptr inbounds ([16 x i8], [16 x i8] addrspace(3)* @x_shared, i32 0, i32 0) to i8*)) #[[ATTR6]]
; CHECK-NEXT: br label [[NEXT:%.*]] ; CHECK-NEXT: br label [[NEXT:%.*]]
; CHECK: next: ; CHECK: next:
; CHECK-NEXT: call void @unknown_no_openmp() ; CHECK-NEXT: call void @unknown_no_openmp()
; CHECK-NEXT: [[B0:%.*]] = icmp eq i32 [[C]], -1 ; CHECK-NEXT: [[B0:%.*]] = icmp eq i32 [[C]], -1
; CHECK-NEXT: br i1 [[B0]], label [[MASTER2:%.*]], label [[EXIT]] ; CHECK-NEXT: br i1 [[B0]], label [[MASTER2:%.*]], label [[EXIT]]
; CHECK: master2: ; CHECK: master2:
; CHECK-NEXT: call void @use.internalized(i8* nofree addrspacecast (i8 addrspace(3)* getelementptr inbounds ([4 x i8], [4 x i8] addrspace(3)* @y_shared, i32 0, i32 0) to i8*)) #[[ATTR6]] ; CHECK-NEXT: call void @use.internalized(i8* nofree addrspacecast (i8 addrspace(3)* getelementptr inbounds ([4 x i8], [4 x i8] addrspace(3)* @y_shared, i32 0, i32 0) to i8*)) #[[ATTR6]]
; CHECK-NEXT: br label [[EXIT]] ; CHECK-NEXT: br label [[EXIT]]
; CHECK: exit: ; CHECK: exit:
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
; ;
; CHECK-LABEL: define {{[^@]+}}@baz_spmd ; CHECK-LABEL: define {{[^@]+}}@baz_spmd
; CHECK-SAME: () #[[ATTR0]] { ; CHECK-SAME: () #[[ATTR0]] {
; CHECK-NEXT: [[C:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 true, i1 true) ; CHECK-NEXT: [[C:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 true)
; CHECK-NEXT: call void @unknown_no_openmp() ; CHECK-NEXT: call void @unknown_no_openmp()
; CHECK-NEXT: [[C0:%.*]] = icmp eq i32 [[C]], -1 ; CHECK-NEXT: [[C0:%.*]] = icmp eq i32 [[C]], -1
; CHECK-NEXT: br i1 [[C0]], label [[MASTER3:%.*]], label [[EXIT:%.*]] ; CHECK-NEXT: br i1 [[C0]], label [[MASTER3:%.*]], label [[EXIT:%.*]]
; CHECK: master3: ; CHECK: master3:
; CHECK-NEXT: [[Z:%.*]] = call align 4 i8* @__kmpc_alloc_shared(i64 24) #[[ATTR6]], !dbg [[DBG10:![0-9]+]] ; CHECK-NEXT: [[Z:%.*]] = call align 4 i8* @__kmpc_alloc_shared(i64 24) #[[ATTR6]], !dbg [[DBG10:![0-9]+]]
; CHECK-NEXT: call void @use.internalized(i8* nofree [[Z]]) #[[ATTR6]] ; CHECK-NEXT: call void @use.internalized(i8* nofree [[Z]]) #[[ATTR6]]
; CHECK-NEXT: call void @__kmpc_free_shared(i8* [[Z]], i64 24) #[[ATTR6]] ; CHECK-NEXT: call void @__kmpc_free_shared(i8* [[Z]], i64 24) #[[ATTR6]]
; CHECK-NEXT: br label [[EXIT]] ; CHECK-NEXT: br label [[EXIT]]
; CHECK: exit: ; CHECK: exit:
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 true) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
; ;
; CHECK: Function Attrs: nofree norecurse nounwind memory(write) ; CHECK: Function Attrs: nofree norecurse nounwind memory(write)
; CHECK-LABEL: define {{[^@]+}}@use.internalized ; CHECK-LABEL: define {{[^@]+}}@use.internalized
; CHECK-SAME: (i8* nofree [[X:%.*]]) #[[ATTR1:[0-9]+]] { ; CHECK-SAME: (i8* nofree [[X:%.*]]) #[[ATTR1:[0-9]+]] {
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: store i8* [[X]], i8** @S, align 8 ; CHECK-NEXT: store i8* [[X]], i8** @S, align 8
Show All 9 Lines
; ;
; CHECK: Function Attrs: nosync nounwind allocsize(0) memory(read) ; CHECK: Function Attrs: nosync nounwind allocsize(0) memory(read)
; CHECK-LABEL: define {{[^@]+}}@__kmpc_alloc_shared ; CHECK-LABEL: define {{[^@]+}}@__kmpc_alloc_shared
; CHECK-SAME: (i64 [[TMP0:%.*]]) #[[ATTR2:[0-9]+]] { ; CHECK-SAME: (i64 [[TMP0:%.*]]) #[[ATTR2:[0-9]+]] {
; CHECK-NEXT: [[L:%.*]] = load i32, i32* @offset, align 4 ; CHECK-NEXT: [[L:%.*]] = load i32, i32* @offset, align 4
; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, i8* addrspacecast (i8 addrspace(3)* getelementptr inbounds ([1024 x i8], [1024 x i8] addrspace(3)* @stack, i32 0, i32 0) to i8*), i32 [[L]] ; CHECK-NEXT: [[GEP:%.*]] = getelementptr i8, i8* addrspacecast (i8 addrspace(3)* getelementptr inbounds ([1024 x i8], [1024 x i8] addrspace(3)* @stack, i32 0, i32 0) to i8*), i32 [[L]]
; CHECK-NEXT: ret i8* [[GEP]] ; CHECK-NEXT: ret i8* [[GEP]]
; ;
;
; CHECK-LABEL: define {{[^@]+}}@__kmpc_target_init
; CHECK-SAME: (%struct.ident_t* [[TMP0:%.*]], i8 [[TMP1:%.*]], i1 [[TMP2:%.*]]) {
; CHECK-NEXT: ret i32 0
;
;. ;.
; CHECK: attributes #[[ATTR0]] = { "kernel" } ; CHECK: attributes #[[ATTR0]] = { "kernel" }
; CHECK: attributes #[[ATTR1]] = { nofree norecurse nounwind memory(write) } ; CHECK: attributes #[[ATTR1]] = { nofree norecurse nounwind memory(write) }
; CHECK: attributes #[[ATTR2]] = { nosync nounwind allocsize(0) memory(read) } ; CHECK: attributes #[[ATTR2]] = { nosync nounwind allocsize(0) memory(read) }
; CHECK: attributes #[[ATTR3:[0-9]+]] = { nosync nounwind } ; CHECK: attributes #[[ATTR3:[0-9]+]] = { nosync nounwind }
; CHECK: attributes #[[ATTR4:[0-9]+]] = { nocallback nofree nosync nounwind speculatable willreturn memory(none) } ; CHECK: attributes #[[ATTR4:[0-9]+]] = { nocallback nofree nosync nounwind speculatable willreturn memory(none) }
; CHECK: attributes #[[ATTR5:[0-9]+]] = { "llvm.assume"="omp_no_openmp" } ; CHECK: attributes #[[ATTR5:[0-9]+]] = { "llvm.assume"="omp_no_openmp" }
; CHECK: attributes #[[ATTR6]] = { nounwind } ; CHECK: attributes #[[ATTR6]] = { nounwind }
;. ;.
; CHECK: [[META0:![0-9]+]] = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 12.0.0", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, enums: !2, splitDebugInlining: false, nameTableKind: None) ; CHECK: [[META0:![0-9]+]] = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 12.0.0", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, enums: !2, splitDebugInlining: false, nameTableKind: None)
; CHECK: [[META1:![0-9]+]] = !DIFile(filename: "replace_globalization.c", directory: "/tmp/replace_globalization.c") ; CHECK: [[META1:![0-9]+]] = !DIFile(filename: "replace_globalization.c", directory: "/tmp/replace_globalization.c")
; CHECK: [[META2:![0-9]+]] = !{} ; CHECK: [[META2:![0-9]+]] = !{}
; CHECK: [[META3:![0-9]+]] = !{i32 2, !"Debug Info Version", i32 3} ; CHECK: [[META3:![0-9]+]] = !{i32 2, !"Debug Info Version", i32 3}
; CHECK: [[META4:![0-9]+]] = !{i32 1, !"wchar_size", i32 4} ; CHECK: [[META4:![0-9]+]] = !{i32 1, !"wchar_size", i32 4}
; CHECK: [[META5:![0-9]+]] = !{i32 7, !"openmp", i32 50} ; CHECK: [[META5:![0-9]+]] = !{i32 7, !"openmp", i32 50}
; CHECK: [[META6:![0-9]+]] = !{i32 7, !"openmp-device", i32 50} ; CHECK: [[META6:![0-9]+]] = !{i32 7, !"openmp-device", i32 50}
; CHECK: [[META7:![0-9]+]] = !{void ()* @foo, !"kernel", i32 1} ; CHECK: [[META7:![0-9]+]] = !{void ()* @foo, !"kernel", i32 1}
; CHECK: [[META8:![0-9]+]] = !{void ()* @bar, !"kernel", i32 1} ; CHECK: [[META8:![0-9]+]] = !{void ()* @bar, !"kernel", i32 1}
; CHECK: [[META9:![0-9]+]] = !{void ()* @baz_spmd, !"kernel", i32 1} ; CHECK: [[META9:![0-9]+]] = !{void ()* @baz_spmd, !"kernel", i32 1}
; CHECK: [[DBG10]] = !DILocation(line: 5, column: 14, scope: !11) ; CHECK: [[DBG10]] = !DILocation(line: 5, column: 14, scope: !11)
; CHECK: [[META11:![0-9]+]] = distinct !DISubprogram(name: "bar", scope: !1, file: !1, line: 1, type: !12, scopeLine: 1, flags: DIFlagPrototyped, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !0, retainedNodes: !2) ; CHECK: [[META11:![0-9]+]] = distinct !DISubprogram(name: "bar", scope: !1, file: !1, line: 1, type: !12, scopeLine: 1, flags: DIFlagPrototyped, spFlags: DISPFlagDefinition | DISPFlagOptimized, unit: !0, retainedNodes: !2)
; CHECK: [[META12:![0-9]+]] = !DISubroutineType(types: !2) ; CHECK: [[META12:![0-9]+]] = !DISubroutineType(types: !2)
;. ;.
;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
; CHECK-LIMIT: {{.*}}
; CHECK-REMARKS: {{.*}}

llvm/test/Transforms/OpenMP/single_threaded_execution.ll

Show All 9 Lines
@kernel_exec_mode = weak constant i8 1 @kernel_exec_mode = weak constant i8 1
; CHECK-NOT: [openmp-opt] Basic block @kernel entry is executed by a single thread. ; CHECK-NOT: [openmp-opt] Basic block @kernel entry is executed by a single thread.
; CHECK: [openmp-opt] Basic block @kernel if.then is executed by a single thread. ; CHECK: [openmp-opt] Basic block @kernel if.then is executed by a single thread.
; CHECK-NOT: [openmp-opt] Basic block @kernel if.else is executed by a single thread. ; CHECK-NOT: [openmp-opt] Basic block @kernel if.else is executed by a single thread.
; CHECK-NOT: [openmp-opt] Basic block @kernel if.end is executed by a single thread. ; CHECK-NOT: [openmp-opt] Basic block @kernel if.end is executed by a single thread.
define void @kernel() { define void @kernel() {
%call = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @1, i8 1, i1 false, i1 false) %call = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @1, i8 1, i1 false)
%cmp = icmp eq i32 %call, -1 %cmp = icmp eq i32 %call, -1
br i1 %cmp, label %if.then, label %if.else br i1 %cmp, label %if.then, label %if.else
if.then: if.then:
br label %if.end br label %if.end
if.else: if.else:
br label %if.end br label %if.end
if.end: if.end:
call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* null, i8 1)
ret void ret void
} }
; REMARKS: remark: single_threaded_execution.c:1:0: Could not internalize function. Some optimizations may not be possible. ; REMARKS: remark: single_threaded_execution.c:1:0: Could not internalize function. Some optimizations may not be possible.
; REMARKS-NOT: remark: single_threaded_execution.c:1:0: Could not internalize function. Some optimizations may not be possible. ; REMARKS-NOT: remark: single_threaded_execution.c:1:0: Could not internalize function. Some optimizations may not be possible.
; CHECK-NOT: [openmp-opt] Basic block @nvptx entry is executed by a single thread. ; CHECK-NOT: [openmp-opt] Basic block @nvptx entry is executed by a single thread.
; CHECK-DAG: [openmp-opt] Basic block @nvptx if.then is executed by a single thread. ; CHECK-DAG: [openmp-opt] Basic block @nvptx if.then is executed by a single thread.
▲ Show 20 Lines • Show All 66 Lines • ▼ Show 20 Lines
} }
declare i32 @llvm.nvvm.read.ptx.sreg.tid.x() declare i32 @llvm.nvvm.read.ptx.sreg.tid.x()
declare i32 @llvm.amdgcn.workitem.id.x() declare i32 @llvm.amdgcn.workitem.id.x()
declare void @__kmpc_kernel_prepare_parallel(i8*) declare void @__kmpc_kernel_prepare_parallel(i8*)
declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1, i1) declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1)
declare void @__kmpc_target_deinit(%struct.ident_t*, i8, i1) declare void @__kmpc_target_deinit(%struct.ident_t*, i8)
attributes #0 = { cold noinline } attributes #0 = { cold noinline }
!llvm.dbg.cu = !{!0} !llvm.dbg.cu = !{!0}
!llvm.module.flags = !{!3, !4, !5, !6} !llvm.module.flags = !{!3, !4, !5, !6}
!nvvm.annotations = !{!7} !nvvm.annotations = !{!7}
!0 = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 12.0.0", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, enums: !2, splitDebugInlining: false, nameTableKind: None) !0 = distinct !DICompileUnit(language: DW_LANG_C99, file: !1, producer: "clang version 12.0.0", isOptimized: false, runtimeVersion: 0, emissionKind: FullDebug, enums: !2, splitDebugInlining: false, nameTableKind: None)
Show All 10 Lines

llvm/test/Transforms/OpenMP/spmdization.ll

Show First 20 Lines • Show All 191 Lines • ▼ Show 20 Lines
} }
define internal void @__omp_offloading_fd02_2044372e_sequential_loop_l5__debug() { define internal void @__omp_offloading_fd02_2044372e_sequential_loop_l5__debug() {
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_l5__debug ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_l5__debug
; AMDGPU-SAME: () #[[ATTR1:[0-9]+]] { ; AMDGPU-SAME: () #[[ATTR1:[0-9]+]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false, i1 false) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false)
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; AMDGPU: common.ret: ; AMDGPU: common.ret:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4:[0-9]+]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4:[0-9]+]]
; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18:![0-9]+]] ; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18:![0-9]+]]
; AMDGPU-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; AMDGPU-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 false) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2)
; AMDGPU-NEXT: br label [[COMMON_RET]] ; AMDGPU-NEXT: br label [[COMMON_RET]]
; ;
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_l5__debug ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_l5__debug
; NVPTX-SAME: () #[[ATTR1:[0-9]+]] { ; NVPTX-SAME: () #[[ATTR1:[0-9]+]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false, i1 false) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false)
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; NVPTX: common.ret: ; NVPTX: common.ret:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4:[0-9]+]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4:[0-9]+]]
; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18:![0-9]+]] ; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18:![0-9]+]]
; NVPTX-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; NVPTX-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 false) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2)
; NVPTX-NEXT: br label [[COMMON_RET]] ; NVPTX-NEXT: br label [[COMMON_RET]]
; ;
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_l5__debug ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_l5__debug
; AMDGPU-DISABLED-SAME: () #[[ATTR1:[0-9]+]] { ; AMDGPU-DISABLED-SAME: () #[[ATTR1:[0-9]+]] {
; AMDGPU-DISABLED-NEXT: entry: ; AMDGPU-DISABLED-NEXT: entry:
; AMDGPU-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5) ; AMDGPU-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5)
; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; AMDGPU-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; AMDGPU-DISABLED: is_worker_check: ; AMDGPU-DISABLED: is_worker_check:
; AMDGPU-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; AMDGPU-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; AMDGPU-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; AMDGPU-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; AMDGPU-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; AMDGPU-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; AMDGPU-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; AMDGPU-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 28 Lines
; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; AMDGPU-DISABLED: common.ret: ; AMDGPU-DISABLED: common.ret:
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; AMDGPU-DISABLED: user_code.entry: ; AMDGPU-DISABLED: user_code.entry:
; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4:[0-9]+]] ; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4:[0-9]+]]
; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18:![0-9]+]] ; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18:![0-9]+]]
; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-DISABLED-NEXT: br label [[COMMON_RET]] ; AMDGPU-DISABLED-NEXT: br label [[COMMON_RET]]
; ;
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_l5__debug ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_l5__debug
; NVPTX-DISABLED-SAME: () #[[ATTR1:[0-9]+]] { ; NVPTX-DISABLED-SAME: () #[[ATTR1:[0-9]+]] {
; NVPTX-DISABLED-NEXT: entry: ; NVPTX-DISABLED-NEXT: entry:
; NVPTX-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8 ; NVPTX-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8
; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; NVPTX-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; NVPTX-DISABLED: is_worker_check: ; NVPTX-DISABLED: is_worker_check:
; NVPTX-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; NVPTX-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; NVPTX-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; NVPTX-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; NVPTX-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; NVPTX-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; NVPTX-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; NVPTX-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 27 Lines
; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; NVPTX-DISABLED: common.ret: ; NVPTX-DISABLED: common.ret:
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; NVPTX-DISABLED: user_code.entry: ; NVPTX-DISABLED: user_code.entry:
; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4:[0-9]+]] ; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4:[0-9]+]]
; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18:![0-9]+]] ; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18:![0-9]+]]
; NVPTX-DISABLED-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; NVPTX-DISABLED-NEXT: call void @__omp_outlined__(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-DISABLED-NEXT: br label [[COMMON_RET]] ; NVPTX-DISABLED-NEXT: br label [[COMMON_RET]]
; ;
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %common.ret br i1 %exec_user_code, label %user_code.entry, label %common.ret
common.ret: ; preds = %entry, %user_code.entry common.ret: ; preds = %entry, %user_code.entry
ret void ret void
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
store i32 %1, i32* %.threadid_temp., align 4, !tbaa !18 store i32 %1, i32* %.threadid_temp., align 4, !tbaa !18
call void @__omp_outlined__(i32* %.threadid_temp., i32* %.zero.addr) #6 call void @__omp_outlined__(i32* %.threadid_temp., i32* %.zero.addr) #6
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
br label %common.ret br label %common.ret
} }
; Function Attrs: alwaysinline convergent norecurse nounwind ; Function Attrs: alwaysinline convergent norecurse nounwind
define internal void @__omp_outlined__(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__ ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
▲ Show 20 Lines • Show All 177 Lines • ▼ Show 20 Lines
; Function Attrs: alwaysinline convergent norecurse nounwind ; Function Attrs: alwaysinline convergent norecurse nounwind
define weak void @__omp_offloading_fd02_2044372e_sequential_loop_to_stack_var_l20() #0 { define weak void @__omp_offloading_fd02_2044372e_sequential_loop_to_stack_var_l20() #0 {
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_stack_var_l20 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_stack_var_l20
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false, i1 false) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false)
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; AMDGPU: common.ret: ; AMDGPU: common.ret:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]] ; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]]
; AMDGPU-NEXT: call void @__omp_outlined__2(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; AMDGPU-NEXT: call void @__omp_outlined__2(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 false) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2)
; AMDGPU-NEXT: br label [[COMMON_RET]] ; AMDGPU-NEXT: br label [[COMMON_RET]]
; ;
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_stack_var_l20 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_stack_var_l20
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false, i1 false) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false)
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; NVPTX: common.ret: ; NVPTX: common.ret:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]] ; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]]
; NVPTX-NEXT: call void @__omp_outlined__2(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; NVPTX-NEXT: call void @__omp_outlined__2(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 false) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2)
; NVPTX-NEXT: br label [[COMMON_RET]] ; NVPTX-NEXT: br label [[COMMON_RET]]
; ;
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_stack_var_l20 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_stack_var_l20
; AMDGPU-DISABLED-SAME: () #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: () #[[ATTR0]] {
; AMDGPU-DISABLED-NEXT: entry: ; AMDGPU-DISABLED-NEXT: entry:
; AMDGPU-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5) ; AMDGPU-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5)
; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; AMDGPU-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; AMDGPU-DISABLED: is_worker_check: ; AMDGPU-DISABLED: is_worker_check:
; AMDGPU-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; AMDGPU-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; AMDGPU-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; AMDGPU-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; AMDGPU-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; AMDGPU-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; AMDGPU-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; AMDGPU-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 28 Lines
; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; AMDGPU-DISABLED: common.ret: ; AMDGPU-DISABLED: common.ret:
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; AMDGPU-DISABLED: user_code.entry: ; AMDGPU-DISABLED: user_code.entry:
; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]] ; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]]
; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__2(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__2(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-DISABLED-NEXT: br label [[COMMON_RET]] ; AMDGPU-DISABLED-NEXT: br label [[COMMON_RET]]
; ;
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_stack_var_l20 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_stack_var_l20
; NVPTX-DISABLED-SAME: () #[[ATTR0]] { ; NVPTX-DISABLED-SAME: () #[[ATTR0]] {
; NVPTX-DISABLED-NEXT: entry: ; NVPTX-DISABLED-NEXT: entry:
; NVPTX-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8 ; NVPTX-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8
; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; NVPTX-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; NVPTX-DISABLED: is_worker_check: ; NVPTX-DISABLED: is_worker_check:
; NVPTX-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; NVPTX-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; NVPTX-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; NVPTX-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; NVPTX-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; NVPTX-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; NVPTX-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; NVPTX-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 27 Lines
; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; NVPTX-DISABLED: common.ret: ; NVPTX-DISABLED: common.ret:
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; NVPTX-DISABLED: user_code.entry: ; NVPTX-DISABLED: user_code.entry:
; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]] ; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]]
; NVPTX-DISABLED-NEXT: call void @__omp_outlined__2(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; NVPTX-DISABLED-NEXT: call void @__omp_outlined__2(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-DISABLED-NEXT: br label [[COMMON_RET]] ; NVPTX-DISABLED-NEXT: br label [[COMMON_RET]]
; ;
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %common.ret br i1 %exec_user_code, label %user_code.entry, label %common.ret
common.ret: ; preds = %entry, %user_code.entry common.ret: ; preds = %entry, %user_code.entry
ret void ret void
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
store i32 %1, i32* %.threadid_temp., align 4, !tbaa !18 store i32 %1, i32* %.threadid_temp., align 4, !tbaa !18
call void @__omp_outlined__2(i32* %.threadid_temp., i32* %.zero.addr) #6 call void @__omp_outlined__2(i32* %.threadid_temp., i32* %.zero.addr) #6
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
br label %common.ret br label %common.ret
} }
; Function Attrs: alwaysinline convergent norecurse nounwind ; Function Attrs: alwaysinline convergent norecurse nounwind
define internal void @__omp_outlined__2(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__2(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__2 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__2
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
▲ Show 20 Lines • Show All 195 Lines • ▼ Show 20 Lines
; Function Attrs: alwaysinline convergent norecurse nounwind ; Function Attrs: alwaysinline convergent norecurse nounwind
define weak void @__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_l35() #0 { define weak void @__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_l35() #0 {
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_l35 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_l35
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false, i1 false) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false)
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; AMDGPU: common.ret: ; AMDGPU: common.ret:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]] ; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]]
; AMDGPU-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; AMDGPU-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 false) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2)
; AMDGPU-NEXT: br label [[COMMON_RET]] ; AMDGPU-NEXT: br label [[COMMON_RET]]
; ;
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_l35 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_l35
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false, i1 false) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false)
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; NVPTX: common.ret: ; NVPTX: common.ret:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]] ; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]]
; NVPTX-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; NVPTX-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 false) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2)
; NVPTX-NEXT: br label [[COMMON_RET]] ; NVPTX-NEXT: br label [[COMMON_RET]]
; ;
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_l35 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_l35
; AMDGPU-DISABLED-SAME: () #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: () #[[ATTR0]] {
; AMDGPU-DISABLED-NEXT: entry: ; AMDGPU-DISABLED-NEXT: entry:
; AMDGPU-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5) ; AMDGPU-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5)
; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; AMDGPU-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; AMDGPU-DISABLED: is_worker_check: ; AMDGPU-DISABLED: is_worker_check:
; AMDGPU-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; AMDGPU-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; AMDGPU-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; AMDGPU-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; AMDGPU-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; AMDGPU-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; AMDGPU-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; AMDGPU-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 28 Lines
; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; AMDGPU-DISABLED: common.ret: ; AMDGPU-DISABLED: common.ret:
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; AMDGPU-DISABLED: user_code.entry: ; AMDGPU-DISABLED: user_code.entry:
; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]] ; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]]
; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-DISABLED-NEXT: br label [[COMMON_RET]] ; AMDGPU-DISABLED-NEXT: br label [[COMMON_RET]]
; ;
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_l35 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_l35
; NVPTX-DISABLED-SAME: () #[[ATTR0]] { ; NVPTX-DISABLED-SAME: () #[[ATTR0]] {
; NVPTX-DISABLED-NEXT: entry: ; NVPTX-DISABLED-NEXT: entry:
; NVPTX-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8 ; NVPTX-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8
; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; NVPTX-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; NVPTX-DISABLED: is_worker_check: ; NVPTX-DISABLED: is_worker_check:
; NVPTX-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; NVPTX-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; NVPTX-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; NVPTX-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; NVPTX-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; NVPTX-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; NVPTX-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; NVPTX-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 27 Lines
; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; NVPTX-DISABLED: common.ret: ; NVPTX-DISABLED: common.ret:
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; NVPTX-DISABLED: user_code.entry: ; NVPTX-DISABLED: user_code.entry:
; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]] ; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]]
; NVPTX-DISABLED-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; NVPTX-DISABLED-NEXT: call void @__omp_outlined__4(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-DISABLED-NEXT: br label [[COMMON_RET]] ; NVPTX-DISABLED-NEXT: br label [[COMMON_RET]]
; ;
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %common.ret br i1 %exec_user_code, label %user_code.entry, label %common.ret
common.ret: ; preds = %entry, %user_code.entry common.ret: ; preds = %entry, %user_code.entry
ret void ret void
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
store i32 %1, i32* %.threadid_temp., align 4, !tbaa !18 store i32 %1, i32* %.threadid_temp., align 4, !tbaa !18
call void @__omp_outlined__4(i32* %.threadid_temp., i32* %.zero.addr) #6 call void @__omp_outlined__4(i32* %.threadid_temp., i32* %.zero.addr) #6
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
br label %common.ret br label %common.ret
} }
; Function Attrs: alwaysinline convergent norecurse nounwind ; Function Attrs: alwaysinline convergent norecurse nounwind
define internal void @__omp_outlined__4(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__4(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__4 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__4
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
▲ Show 20 Lines • Show All 220 Lines • ▼ Show 20 Lines
; Function Attrs: alwaysinline convergent norecurse nounwind ; Function Attrs: alwaysinline convergent norecurse nounwind
define weak void @__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_guarded_l50() #0 { define weak void @__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_guarded_l50() #0 {
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_guarded_l50 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_guarded_l50
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false, i1 false) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false)
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; AMDGPU: common.ret: ; AMDGPU: common.ret:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]] ; AMDGPU-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]]
; AMDGPU-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; AMDGPU-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 false) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2)
; AMDGPU-NEXT: br label [[COMMON_RET]] ; AMDGPU-NEXT: br label [[COMMON_RET]]
; ;
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_guarded_l50 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_guarded_l50
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false, i1 false) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false)
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; NVPTX: common.ret: ; NVPTX: common.ret:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]] ; NVPTX-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]]
; NVPTX-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; NVPTX-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 false) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2)
; NVPTX-NEXT: br label [[COMMON_RET]] ; NVPTX-NEXT: br label [[COMMON_RET]]
; ;
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_guarded_l50 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_guarded_l50
; AMDGPU-DISABLED-SAME: () #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: () #[[ATTR0]] {
; AMDGPU-DISABLED-NEXT: entry: ; AMDGPU-DISABLED-NEXT: entry:
; AMDGPU-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5) ; AMDGPU-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5)
; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; AMDGPU-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; AMDGPU-DISABLED: is_worker_check: ; AMDGPU-DISABLED: is_worker_check:
; AMDGPU-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; AMDGPU-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; AMDGPU-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; AMDGPU-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; AMDGPU-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; AMDGPU-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; AMDGPU-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; AMDGPU-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 28 Lines
; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; AMDGPU-DISABLED: common.ret: ; AMDGPU-DISABLED: common.ret:
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; AMDGPU-DISABLED: user_code.entry: ; AMDGPU-DISABLED: user_code.entry:
; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]] ; AMDGPU-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]]
; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-DISABLED-NEXT: br label [[COMMON_RET]] ; AMDGPU-DISABLED-NEXT: br label [[COMMON_RET]]
; ;
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_guarded_l50 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_sequential_loop_to_shared_var_guarded_l50
; NVPTX-DISABLED-SAME: () #[[ATTR0]] { ; NVPTX-DISABLED-SAME: () #[[ATTR0]] {
; NVPTX-DISABLED-NEXT: entry: ; NVPTX-DISABLED-NEXT: entry:
; NVPTX-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8 ; NVPTX-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8
; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; NVPTX-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; NVPTX-DISABLED: is_worker_check: ; NVPTX-DISABLED: is_worker_check:
; NVPTX-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; NVPTX-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; NVPTX-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; NVPTX-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; NVPTX-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; NVPTX-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; NVPTX-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; NVPTX-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 27 Lines
; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; NVPTX-DISABLED: common.ret: ; NVPTX-DISABLED: common.ret:
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; NVPTX-DISABLED: user_code.entry: ; NVPTX-DISABLED: user_code.entry:
; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]] ; NVPTX-DISABLED-NEXT: store i32 [[TMP1]], i32* [[DOTTHREADID_TEMP_]], align 4, !tbaa [[TBAA18]]
; NVPTX-DISABLED-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; NVPTX-DISABLED-NEXT: call void @__omp_outlined__6(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-DISABLED-NEXT: br label [[COMMON_RET]] ; NVPTX-DISABLED-NEXT: br label [[COMMON_RET]]
; ;
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %common.ret br i1 %exec_user_code, label %user_code.entry, label %common.ret
common.ret: ; preds = %entry, %user_code.entry common.ret: ; preds = %entry, %user_code.entry
ret void ret void
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
store i32 %1, i32* %.threadid_temp., align 4, !tbaa !18 store i32 %1, i32* %.threadid_temp., align 4, !tbaa !18
call void @__omp_outlined__6(i32* %.threadid_temp., i32* %.zero.addr) #6 call void @__omp_outlined__6(i32* %.threadid_temp., i32* %.zero.addr) #6
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
br label %common.ret br label %common.ret
} }
; Function Attrs: alwaysinline convergent norecurse nounwind ; Function Attrs: alwaysinline convergent norecurse nounwind
define internal void @__omp_outlined__6(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__6(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__6 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__6
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
▲ Show 20 Lines • Show All 256 Lines • ▼ Show 20 Lines
; Function Attrs: alwaysinline convergent norecurse nounwind ; Function Attrs: alwaysinline convergent norecurse nounwind
define weak void @__omp_offloading_fd02_2044372e_do_not_spmdize_target_l65() #0 { define weak void @__omp_offloading_fd02_2044372e_do_not_spmdize_target_l65() #0 {
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_do_not_spmdize_target_l65 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_do_not_spmdize_target_l65
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5) ; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5)
; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; AMDGPU: is_worker_check: ; AMDGPU: is_worker_check:
; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 21 Lines
; AMDGPU: thread.user_code.check: ; AMDGPU: thread.user_code.check:
; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; AMDGPU-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; AMDGPU: common.ret: ; AMDGPU: common.ret:
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; AMDGPU-NEXT: call void @__omp_outlined__8(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; AMDGPU-NEXT: call void @__omp_outlined__8(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: br label [[COMMON_RET]] ; AMDGPU-NEXT: br label [[COMMON_RET]]
; ;
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_do_not_spmdize_target_l65 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_do_not_spmdize_target_l65
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8 ; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8
; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; NVPTX: is_worker_check: ; NVPTX: is_worker_check:
; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 20 Lines
; NVPTX: thread.user_code.check: ; NVPTX: thread.user_code.check:
; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; NVPTX-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; NVPTX: common.ret: ; NVPTX: common.ret:
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; NVPTX-NEXT: call void @__omp_outlined__8(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; NVPTX-NEXT: call void @__omp_outlined__8(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: br label [[COMMON_RET]] ; NVPTX-NEXT: br label [[COMMON_RET]]
; ;
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_do_not_spmdize_target_l65 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_do_not_spmdize_target_l65
; AMDGPU-DISABLED-SAME: () #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: () #[[ATTR0]] {
; AMDGPU-DISABLED-NEXT: entry: ; AMDGPU-DISABLED-NEXT: entry:
; AMDGPU-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5) ; AMDGPU-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5)
; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; AMDGPU-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; AMDGPU-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; AMDGPU-DISABLED: is_worker_check: ; AMDGPU-DISABLED: is_worker_check:
; AMDGPU-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; AMDGPU-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; AMDGPU-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; AMDGPU-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; AMDGPU-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; AMDGPU-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; AMDGPU-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; AMDGPU-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 21 Lines
; AMDGPU-DISABLED: thread.user_code.check: ; AMDGPU-DISABLED: thread.user_code.check:
; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; AMDGPU-DISABLED: common.ret: ; AMDGPU-DISABLED: common.ret:
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; AMDGPU-DISABLED: user_code.entry: ; AMDGPU-DISABLED: user_code.entry:
; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__8(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; AMDGPU-DISABLED-NEXT: call void @__omp_outlined__8(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-DISABLED-NEXT: br label [[COMMON_RET]] ; AMDGPU-DISABLED-NEXT: br label [[COMMON_RET]]
; ;
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_do_not_spmdize_target_l65 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_do_not_spmdize_target_l65
; NVPTX-DISABLED-SAME: () #[[ATTR0]] { ; NVPTX-DISABLED-SAME: () #[[ATTR0]] {
; NVPTX-DISABLED-NEXT: entry: ; NVPTX-DISABLED-NEXT: entry:
; NVPTX-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8 ; NVPTX-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8
; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTZERO_ADDR:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4 ; NVPTX-DISABLED-NEXT: [[DOTTHREADID_TEMP_:%.*]] = alloca i32, align 4
; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; NVPTX-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; NVPTX-DISABLED: is_worker_check: ; NVPTX-DISABLED: is_worker_check:
; NVPTX-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; NVPTX-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; NVPTX-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; NVPTX-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; NVPTX-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; NVPTX-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; NVPTX-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; NVPTX-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 20 Lines
; NVPTX-DISABLED: thread.user_code.check: ; NVPTX-DISABLED: thread.user_code.check:
; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; NVPTX-DISABLED: common.ret: ; NVPTX-DISABLED: common.ret:
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; NVPTX-DISABLED: user_code.entry: ; NVPTX-DISABLED: user_code.entry:
; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; NVPTX-DISABLED-NEXT: call void @__omp_outlined__8(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]] ; NVPTX-DISABLED-NEXT: call void @__omp_outlined__8(i32* [[DOTTHREADID_TEMP_]], i32* [[DOTZERO_ADDR]]) #[[ATTR4]]
; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-DISABLED-NEXT: br label [[COMMON_RET]] ; NVPTX-DISABLED-NEXT: br label [[COMMON_RET]]
; ;
entry: entry:
%.zero.addr = alloca i32, align 4 %.zero.addr = alloca i32, align 4
%.threadid_temp. = alloca i32, align 4 %.threadid_temp. = alloca i32, align 4
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %common.ret br i1 %exec_user_code, label %user_code.entry, label %common.ret
common.ret: ; preds = %entry, %user_code.entry common.ret: ; preds = %entry, %user_code.entry
ret void ret void
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
store i32 0, i32* %.zero.addr, align 4 store i32 0, i32* %.zero.addr, align 4
store i32 %1, i32* %.threadid_temp., align 4, !tbaa !18 store i32 %1, i32* %.threadid_temp., align 4, !tbaa !18
call void @__omp_outlined__8(i32* %.threadid_temp., i32* %.zero.addr) #6 call void @__omp_outlined__8(i32* %.threadid_temp., i32* %.zero.addr) #6
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
br label %common.ret br label %common.ret
} }
; Function Attrs: alwaysinline convergent norecurse nounwind ; Function Attrs: alwaysinline convergent norecurse nounwind
define internal void @__omp_outlined__8(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__8(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__8 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__8
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
Show All 25 Lines
; Function Attrs: alwaysinline convergent norecurse nounwind ; Function Attrs: alwaysinline convergent norecurse nounwind
define weak void @__omp_offloading_fd02_2044372e_do_not_spmdize_task_l74() #0 { define weak void @__omp_offloading_fd02_2044372e_do_not_spmdize_task_l74() #0 {
; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_do_not_spmdize_task_l74 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_do_not_spmdize_task_l74
; AMDGPU-SAME: () #[[ATTR0]] { ; AMDGPU-SAME: () #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5) ; AMDGPU-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5)
; AMDGPU-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x i8*], align 8 ; AMDGPU-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x i8*], align 8
; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; AMDGPU-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; AMDGPU-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; AMDGPU: is_worker_check: ; AMDGPU: is_worker_check:
; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; AMDGPU-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; AMDGPU-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; AMDGPU-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; AMDGPU-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; AMDGPU-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 31 Lines
; AMDGPU-NEXT: ret void ; AMDGPU-NEXT: ret void
; AMDGPU: user_code.entry: ; AMDGPU: user_code.entry:
; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; AMDGPU-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; AMDGPU-NEXT: [[TMP2:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i64 40, i64 0, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @"_omp_task_entry$" to i32 (i32, i8*)*)) #[[ATTR4]] ; AMDGPU-NEXT: [[TMP2:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i64 40, i64 0, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @"_omp_task_entry$" to i32 (i32, i8*)*)) #[[ATTR4]]
; AMDGPU-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.kmp_task_t_with_privates* ; AMDGPU-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.kmp_task_t_with_privates*
; AMDGPU-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i8* [[TMP2]]) #[[ATTR4]] ; AMDGPU-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i8* [[TMP2]]) #[[ATTR4]]
; AMDGPU-NEXT: [[TMP5:%.*]] = bitcast [0 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8** ; AMDGPU-NEXT: [[TMP5:%.*]] = bitcast [0 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8**
; AMDGPU-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__9 to i8*), i8* @__omp_outlined__9_wrapper.ID, i8** [[TMP5]], i64 0) ; AMDGPU-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__9 to i8*), i8* @__omp_outlined__9_wrapper.ID, i8** [[TMP5]], i64 0)
; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-NEXT: br label [[COMMON_RET]] ; AMDGPU-NEXT: br label [[COMMON_RET]]
; ;
; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_do_not_spmdize_task_l74 ; NVPTX-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_do_not_spmdize_task_l74
; NVPTX-SAME: () #[[ATTR0]] { ; NVPTX-SAME: () #[[ATTR0]] {
; NVPTX-NEXT: entry: ; NVPTX-NEXT: entry:
; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8 ; NVPTX-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8
; NVPTX-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x i8*], align 8 ; NVPTX-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x i8*], align 8
; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; NVPTX-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; NVPTX-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; NVPTX: is_worker_check: ; NVPTX: is_worker_check:
; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; NVPTX-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; NVPTX-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; NVPTX-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; NVPTX-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; NVPTX-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 30 Lines
; NVPTX-NEXT: ret void ; NVPTX-NEXT: ret void
; NVPTX: user_code.entry: ; NVPTX: user_code.entry:
; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; NVPTX-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; NVPTX-NEXT: [[TMP2:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i64 40, i64 0, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @"_omp_task_entry$" to i32 (i32, i8*)*)) #[[ATTR4]] ; NVPTX-NEXT: [[TMP2:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i64 40, i64 0, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @"_omp_task_entry$" to i32 (i32, i8*)*)) #[[ATTR4]]
; NVPTX-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.kmp_task_t_with_privates* ; NVPTX-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.kmp_task_t_with_privates*
; NVPTX-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i8* [[TMP2]]) #[[ATTR4]] ; NVPTX-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i8* [[TMP2]]) #[[ATTR4]]
; NVPTX-NEXT: [[TMP5:%.*]] = bitcast [0 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8** ; NVPTX-NEXT: [[TMP5:%.*]] = bitcast [0 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8**
; NVPTX-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__9 to i8*), i8* @__omp_outlined__9_wrapper.ID, i8** [[TMP5]], i64 0) ; NVPTX-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__9 to i8*), i8* @__omp_outlined__9_wrapper.ID, i8** [[TMP5]], i64 0)
; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-NEXT: br label [[COMMON_RET]] ; NVPTX-NEXT: br label [[COMMON_RET]]
; ;
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_do_not_spmdize_task_l74 ; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_do_not_spmdize_task_l74
; AMDGPU-DISABLED-SAME: () #[[ATTR0]] { ; AMDGPU-DISABLED-SAME: () #[[ATTR0]] {
; AMDGPU-DISABLED-NEXT: entry: ; AMDGPU-DISABLED-NEXT: entry:
; AMDGPU-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5) ; AMDGPU-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8, addrspace(5)
; AMDGPU-DISABLED-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x i8*], align 8 ; AMDGPU-DISABLED-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x i8*], align 8
; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; AMDGPU-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; AMDGPU-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; AMDGPU-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; AMDGPU-DISABLED: is_worker_check: ; AMDGPU-DISABLED: is_worker_check:
; AMDGPU-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; AMDGPU-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; AMDGPU-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; AMDGPU-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; AMDGPU-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; AMDGPU-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; AMDGPU-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; AMDGPU-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; AMDGPU-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 31 Lines
; AMDGPU-DISABLED-NEXT: ret void ; AMDGPU-DISABLED-NEXT: ret void
; AMDGPU-DISABLED: user_code.entry: ; AMDGPU-DISABLED: user_code.entry:
; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; AMDGPU-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; AMDGPU-DISABLED-NEXT: [[TMP2:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i64 40, i64 0, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @"_omp_task_entry$" to i32 (i32, i8*)*)) #[[ATTR4]] ; AMDGPU-DISABLED-NEXT: [[TMP2:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i64 40, i64 0, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @"_omp_task_entry$" to i32 (i32, i8*)*)) #[[ATTR4]]
; AMDGPU-DISABLED-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.kmp_task_t_with_privates* ; AMDGPU-DISABLED-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.kmp_task_t_with_privates*
; AMDGPU-DISABLED-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i8* [[TMP2]]) #[[ATTR4]] ; AMDGPU-DISABLED-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i8* [[TMP2]]) #[[ATTR4]]
; AMDGPU-DISABLED-NEXT: [[TMP5:%.*]] = bitcast [0 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8** ; AMDGPU-DISABLED-NEXT: [[TMP5:%.*]] = bitcast [0 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8**
; AMDGPU-DISABLED-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__9 to i8*), i8* @__omp_outlined__9_wrapper.ID, i8** [[TMP5]], i64 0) ; AMDGPU-DISABLED-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__9 to i8*), i8* @__omp_outlined__9_wrapper.ID, i8** [[TMP5]], i64 0)
; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; AMDGPU-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; AMDGPU-DISABLED-NEXT: br label [[COMMON_RET]] ; AMDGPU-DISABLED-NEXT: br label [[COMMON_RET]]
; ;
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_do_not_spmdize_task_l74 ; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_fd02_2044372e_do_not_spmdize_task_l74
; NVPTX-DISABLED-SAME: () #[[ATTR0]] { ; NVPTX-DISABLED-SAME: () #[[ATTR0]] {
; NVPTX-DISABLED-NEXT: entry: ; NVPTX-DISABLED-NEXT: entry:
; NVPTX-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8 ; NVPTX-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8
; NVPTX-DISABLED-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x i8*], align 8 ; NVPTX-DISABLED-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x i8*], align 8
; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; NVPTX-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; NVPTX-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; NVPTX-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; NVPTX-DISABLED: is_worker_check: ; NVPTX-DISABLED: is_worker_check:
; NVPTX-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; NVPTX-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; NVPTX-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; NVPTX-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; NVPTX-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; NVPTX-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; NVPTX-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; NVPTX-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; NVPTX-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 30 Lines
; NVPTX-DISABLED-NEXT: ret void ; NVPTX-DISABLED-NEXT: ret void
; NVPTX-DISABLED: user_code.entry: ; NVPTX-DISABLED: user_code.entry:
; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]] ; NVPTX-DISABLED-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB1]]) #[[ATTR4]]
; NVPTX-DISABLED-NEXT: [[TMP2:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i64 40, i64 0, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @"_omp_task_entry$" to i32 (i32, i8*)*)) #[[ATTR4]] ; NVPTX-DISABLED-NEXT: [[TMP2:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i64 40, i64 0, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @"_omp_task_entry$" to i32 (i32, i8*)*)) #[[ATTR4]]
; NVPTX-DISABLED-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.kmp_task_t_with_privates* ; NVPTX-DISABLED-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to %struct.kmp_task_t_with_privates*
; NVPTX-DISABLED-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i8* [[TMP2]]) #[[ATTR4]] ; NVPTX-DISABLED-NEXT: [[TMP4:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i8* [[TMP2]]) #[[ATTR4]]
; NVPTX-DISABLED-NEXT: [[TMP5:%.*]] = bitcast [0 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8** ; NVPTX-DISABLED-NEXT: [[TMP5:%.*]] = bitcast [0 x i8*]* [[CAPTURED_VARS_ADDRS]] to i8**
; NVPTX-DISABLED-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__9 to i8*), i8* @__omp_outlined__9_wrapper.ID, i8** [[TMP5]], i64 0) ; NVPTX-DISABLED-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__9 to i8*), i8* @__omp_outlined__9_wrapper.ID, i8** [[TMP5]], i64 0)
; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; NVPTX-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; NVPTX-DISABLED-NEXT: br label [[COMMON_RET]] ; NVPTX-DISABLED-NEXT: br label [[COMMON_RET]]
; ;
entry: entry:
%captured_vars_addrs = alloca [0 x i8*], align 8 %captured_vars_addrs = alloca [0 x i8*], align 8
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %common.ret br i1 %exec_user_code, label %user_code.entry, label %common.ret
common.ret: ; preds = %entry, %user_code.entry common.ret: ; preds = %entry, %user_code.entry
ret void ret void
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* @1)
%2 = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @1, i32 %1, i32 1, i64 40, i64 0, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @"_omp_task_entry$" to i32 (i32, i8*)*)) %2 = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @1, i32 %1, i32 1, i64 40, i64 0, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @"_omp_task_entry$" to i32 (i32, i8*)*))
%3 = bitcast i8* %2 to %struct.kmp_task_t_with_privates* %3 = bitcast i8* %2 to %struct.kmp_task_t_with_privates*
%4 = call i32 @__kmpc_omp_task(%struct.ident_t* @1, i32 %1, i8* %2) %4 = call i32 @__kmpc_omp_task(%struct.ident_t* @1, i32 %1, i8* %2)
%5 = bitcast [0 x i8*]* %captured_vars_addrs to i8** %5 = bitcast [0 x i8*]* %captured_vars_addrs to i8**
call void @__kmpc_parallel_51(%struct.ident_t* @1, i32 %1, i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__9 to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined__9_wrapper to i8*), i8** %5, i64 0) call void @__kmpc_parallel_51(%struct.ident_t* @1, i32 %1, i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__9 to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined__9_wrapper to i8*), i8** %5, i64 0)
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
br label %common.ret br label %common.ret
} }
; Function Attrs: alwaysinline convergent nounwind ; Function Attrs: alwaysinline convergent nounwind
define internal void @.omp_outlined.(i32 %.global_tid., i32* noalias %.part_id., i8* noalias %.privates., void (i8*, ...)* noalias %.copy_fn., i8* %.task_t., %struct.anon* noalias %__context) #9 { define internal void @.omp_outlined.(i32 %.global_tid., i32* noalias %.part_id., i8* noalias %.privates., void (i8*, ...)* noalias %.copy_fn., i8* %.task_t., %struct.anon* noalias %__context) #9 {
; AMDGPU-LABEL: define {{[^@]+}}@.omp_outlined. ; AMDGPU-LABEL: define {{[^@]+}}@.omp_outlined.
; AMDGPU-SAME: (i32 [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTPART_ID_:%.*]], i8* noalias [[DOTPRIVATES_:%.*]], void (i8*, ...)* noalias [[DOTCOPY_FN_:%.*]], i8* [[DOTTASK_T_:%.*]], %struct.anon* noalias [[__CONTEXT:%.*]]) #[[ATTR3:[0-9]+]] { ; AMDGPU-SAME: (i32 [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTPART_ID_:%.*]], i8* noalias [[DOTPRIVATES_:%.*]], void (i8*, ...)* noalias [[DOTCOPY_FN_:%.*]], i8* [[DOTTASK_T_:%.*]], %struct.anon* noalias [[__CONTEXT:%.*]]) #[[ATTR3:[0-9]+]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
▲ Show 20 Lines • Show All 54 Lines • ▼ Show 20 Lines
; Function Attrs: convergent ; Function Attrs: convergent
declare void @unknown() #2 declare void @unknown() #2
declare void @unknowni32p(i32*) #2 declare void @unknowni32p(i32*) #2
; Function Attrs: argmemonly mustprogress nofree nosync nounwind willreturn ; Function Attrs: argmemonly mustprogress nofree nosync nounwind willreturn
declare void @llvm.lifetime.start.p0i8(i64 immarg, i8* nocapture) #1 declare void @llvm.lifetime.start.p0i8(i64 immarg, i8* nocapture) #1
; Make it a weak definition so we will apply custom state machine rewriting but can't use the body in the reasoning. ; Make it a weak definition so we will apply custom state machine rewriting but can't use the body in the reasoning.
define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1, i1) { define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1) {
; AMDGPU-LABEL: define {{[^@]+}}@__kmpc_target_init
; AMDGPU-SAME: (%struct.ident_t* [[TMP0:%.*]], i8 [[TMP1:%.*]], i1 [[TMP2:%.*]]) {
; AMDGPU-NEXT: ret i32 0
;
; NVPTX-LABEL: define {{[^@]+}}@__kmpc_target_init
; NVPTX-SAME: (%struct.ident_t* [[TMP0:%.*]], i8 [[TMP1:%.*]], i1 [[TMP2:%.*]]) {
; NVPTX-NEXT: ret i32 0
;
; AMDGPU-DISABLED-LABEL: define {{[^@]+}}@__kmpc_target_init
; AMDGPU-DISABLED-SAME: (%struct.ident_t* [[TMP0:%.*]], i8 [[TMP1:%.*]], i1 [[TMP2:%.*]]) {
; AMDGPU-DISABLED-NEXT: ret i32 0
;
; NVPTX-DISABLED-LABEL: define {{[^@]+}}@__kmpc_target_init
; NVPTX-DISABLED-SAME: (%struct.ident_t* [[TMP0:%.*]], i8 [[TMP1:%.*]], i1 [[TMP2:%.*]]) {
; NVPTX-DISABLED-NEXT: ret i32 0
;
ret i32 0 ret i32 0
} }
declare void @__kmpc_get_shared_variables(i8***) declare void @__kmpc_get_shared_variables(i8***)
; Function Attrs: alwaysinline ; Function Attrs: alwaysinline
declare void @__kmpc_parallel_51(%struct.ident_t*, i32, i32, i32, i32, i8*, i8*, i8**, i64) #4 declare void @__kmpc_parallel_51(%struct.ident_t*, i32, i32, i32, i32, i8*, i8*, i8**, i64) #4
; Function Attrs: argmemonly mustprogress nofree nosync nounwind willreturn ; Function Attrs: argmemonly mustprogress nofree nosync nounwind willreturn
declare void @llvm.lifetime.end.p0i8(i64 immarg, i8* nocapture) #1 declare void @llvm.lifetime.end.p0i8(i64 immarg, i8* nocapture) #1
; Function Attrs: convergent ; Function Attrs: convergent
declare void @spmd_amenable() #5 declare void @spmd_amenable() #5
; Function Attrs: nounwind ; Function Attrs: nounwind
declare i32 @__kmpc_global_thread_num(%struct.ident_t*) #6 declare i32 @__kmpc_global_thread_num(%struct.ident_t*) #6
declare void @__kmpc_target_deinit(%struct.ident_t*, i8, i1) declare void @__kmpc_target_deinit(%struct.ident_t*, i8)
; Function Attrs: alwaysinline convergent norecurse nounwind ; Function Attrs: alwaysinline convergent norecurse nounwind
define internal void @__omp_outlined__9(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 { define internal void @__omp_outlined__9(i32* noalias %.global_tid., i32* noalias %.bound_tid.) #0 {
; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__9 ; AMDGPU-LABEL: define {{[^@]+}}@__omp_outlined__9
; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] { ; AMDGPU-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]]) #[[ATTR0]] {
; AMDGPU-NEXT: entry: ; AMDGPU-NEXT: entry:
; AMDGPU-NEXT: call void @unknown() #[[ATTR8]] ; AMDGPU-NEXT: call void @unknown() #[[ATTR8]]
▲ Show 20 Lines • Show All 320 Lines • Show Last 20 Lines

llvm/test/Transforms/OpenMP/spmdization_assumes.ll

Show All 26 Lines
; CHECK: @[[LLVM_COMPILER_USED:[a-zA-Z0-9_$"\\.-]+]] = appending global [1 x i8*] [i8* @__omp_offloading_fd02_404433c2_main_l5_exec_mode], section "llvm.metadata" ; CHECK: @[[LLVM_COMPILER_USED:[a-zA-Z0-9_$"\\.-]+]] = appending global [1 x i8*] [i8* @__omp_offloading_fd02_404433c2_main_l5_exec_mode], section "llvm.metadata"
; CHECK: @[[GLOB2:[0-9]+]] = private unnamed_addr constant [[STRUCT_IDENT_T:%.*]] { i32 0, i32 2, i32 0, i32 22, i8* getelementptr inbounds ([23 x i8], [23 x i8]* @[[GLOB0]], i32 0, i32 0) }, align 8 ; CHECK: @[[GLOB2:[0-9]+]] = private unnamed_addr constant [[STRUCT_IDENT_T:%.*]] { i32 0, i32 2, i32 0, i32 22, i8* getelementptr inbounds ([23 x i8], [23 x i8]* @[[GLOB0]], i32 0, i32 0) }, align 8
;. ;.
define weak void @__omp_offloading_fd02_404433c2_main_l5(double* nonnull align 8 dereferenceable(8) %x) local_unnamed_addr #0 { define weak void @__omp_offloading_fd02_404433c2_main_l5(double* nonnull align 8 dereferenceable(8) %x) local_unnamed_addr #0 {
; CHECK-LABEL: define {{[^@]+}}@__omp_offloading_fd02_404433c2_main_l5 ; CHECK-LABEL: define {{[^@]+}}@__omp_offloading_fd02_404433c2_main_l5
; CHECK-SAME: (double* nonnull align 8 dereferenceable(8) [[X:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] { ; CHECK-SAME: (double* nonnull align 8 dereferenceable(8) [[X:%.*]]) local_unnamed_addr #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x i8*], align 8 ; CHECK-NEXT: [[CAPTURED_VARS_ADDRS:%.*]] = alloca [0 x i8*], align 8
; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @[[GLOB1]], i8 2, i1 false, i1 false) #[[ATTR3:[0-9]+]] ; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @[[GLOB1]], i8 2, i1 false) #[[ATTR3:[0-9]+]]
; CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; CHECK-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]] ; CHECK-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[COMMON_RET:%.*]]
; CHECK: common.ret: ; CHECK: common.ret:
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; CHECK: user_code.entry: ; CHECK: user_code.entry:
; CHECK-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @[[GLOB1]]) #[[ATTR3]] ; CHECK-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @[[GLOB1]]) #[[ATTR3]]
; CHECK-NEXT: [[CALL_I:%.*]] = call double @__nv_sin(double 0x400921FB54442D18) #[[ATTR7:[0-9]+]] ; CHECK-NEXT: [[CALL_I:%.*]] = call double @__nv_sin(double 0x400921FB54442D18) #[[ATTR7:[0-9]+]]
; CHECK-NEXT: br label [[REGION_CHECK_TID:%.*]] ; CHECK-NEXT: br label [[REGION_CHECK_TID:%.*]]
; CHECK: region.check.tid: ; CHECK: region.check.tid:
; CHECK-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_get_hardware_thread_id_in_block() ; CHECK-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_get_hardware_thread_id_in_block()
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i32 [[TMP2]], 0 ; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i32 [[TMP2]], 0
; CHECK-NEXT: br i1 [[TMP3]], label [[REGION_GUARDED:%.*]], label [[REGION_BARRIER:%.*]] ; CHECK-NEXT: br i1 [[TMP3]], label [[REGION_GUARDED:%.*]], label [[REGION_BARRIER:%.*]]
; CHECK: region.guarded: ; CHECK: region.guarded:
; CHECK-NEXT: store double [[CALL_I]], double* [[X]], align 8, !tbaa [[TBAA8:![0-9]+]] ; CHECK-NEXT: store double [[CALL_I]], double* [[X]], align 8, !tbaa [[TBAA8:![0-9]+]]
; CHECK-NEXT: br label [[REGION_GUARDED_END:%.*]] ; CHECK-NEXT: br label [[REGION_GUARDED_END:%.*]]
; CHECK: region.guarded.end: ; CHECK: region.guarded.end:
; CHECK-NEXT: br label [[REGION_BARRIER]] ; CHECK-NEXT: br label [[REGION_BARRIER]]
; CHECK: region.barrier: ; CHECK: region.barrier:
; CHECK-NEXT: call void @__kmpc_barrier_simple_spmd(%struct.ident_t* @[[GLOB2]], i32 [[TMP2]]) ; CHECK-NEXT: call void @__kmpc_barrier_simple_spmd(%struct.ident_t* @[[GLOB2]], i32 [[TMP2]])
; CHECK-NEXT: br label [[REGION_EXIT:%.*]] ; CHECK-NEXT: br label [[REGION_EXIT:%.*]]
; CHECK: region.exit: ; CHECK: region.exit:
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds [0 x i8*], [0 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 0 ; CHECK-NEXT: [[TMP4:%.*]] = getelementptr inbounds [0 x i8*], [0 x i8*]* [[CAPTURED_VARS_ADDRS]], i64 0, i64 0
; CHECK-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* nonnull @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__ to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined___wrapper to i8*), i8** nonnull [[TMP4]], i64 0) #[[ATTR3]] ; CHECK-NEXT: call void @__kmpc_parallel_51(%struct.ident_t* nonnull @[[GLOB1]], i32 [[TMP1]], i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__ to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined___wrapper to i8*), i8** nonnull [[TMP4]], i64 0) #[[ATTR3]]
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* nonnull @[[GLOB1]], i8 2, i1 false) #[[ATTR3]] ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* nonnull @[[GLOB1]], i8 2) #[[ATTR3]]
; CHECK-NEXT: br label [[COMMON_RET]] ; CHECK-NEXT: br label [[COMMON_RET]]
; ;
entry: entry:
%captured_vars_addrs = alloca [0 x i8*], align 8 %captured_vars_addrs = alloca [0 x i8*], align 8
%0 = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @1, i8 1, i1 true, i1 true) #3 %0 = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @1, i8 1, i1 true) #3
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %common.ret br i1 %exec_user_code, label %user_code.entry, label %common.ret
common.ret: ; preds = %entry, %user_code.entry common.ret: ; preds = %entry, %user_code.entry
ret void ret void
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1)
%call.i = call double @__nv_sin(double 0x400921FB54442D18) #6 %call.i = call double @__nv_sin(double 0x400921FB54442D18) #6
store double %call.i, double* %x, align 8, !tbaa !8 store double %call.i, double* %x, align 8, !tbaa !8
%2 = getelementptr inbounds [0 x i8*], [0 x i8*]* %captured_vars_addrs, i64 0, i64 0 %2 = getelementptr inbounds [0 x i8*], [0 x i8*]* %captured_vars_addrs, i64 0, i64 0
call void @__kmpc_parallel_51(%struct.ident_t* nonnull @1, i32 %1, i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__ to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined___wrapper to i8*), i8** nonnull %2, i64 0) #3 call void @__kmpc_parallel_51(%struct.ident_t* nonnull @1, i32 %1, i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__ to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined___wrapper to i8*), i8** nonnull %2, i64 0) #3
call void @__kmpc_target_deinit(%struct.ident_t* nonnull @1, i8 1, i1 true) #3 call void @__kmpc_target_deinit(%struct.ident_t* nonnull @1, i8 1) #3
br label %common.ret br label %common.ret
} }
declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1, i1) local_unnamed_addr declare i32 @__kmpc_target_init(%struct.ident_t*, i8, i1) local_unnamed_addr
; Function Attrs: alwaysinline mustprogress nofree norecurse nosync nounwind readnone willreturn ; Function Attrs: alwaysinline mustprogress nofree norecurse nosync nounwind readnone willreturn
define internal void @__omp_outlined__(i32* noalias nocapture %.global_tid., i32* noalias nocapture %.bound_tid.) #1 { define internal void @__omp_outlined__(i32* noalias nocapture %.global_tid., i32* noalias nocapture %.bound_tid.) #1 {
; CHECK-LABEL: define {{[^@]+}}@__omp_outlined__ ; CHECK-LABEL: define {{[^@]+}}@__omp_outlined__
; CHECK-SAME: (i32* noalias nocapture [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] { ; CHECK-SAME: (i32* noalias nocapture [[DOTGLOBAL_TID_:%.*]], i32* noalias nocapture [[DOTBOUND_TID_:%.*]]) #[[ATTR1:[0-9]+]] {
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
Show All 19 Lines
declare void @__kmpc_get_shared_variables(i8***) local_unnamed_addr declare void @__kmpc_get_shared_variables(i8***) local_unnamed_addr
; Function Attrs: nounwind ; Function Attrs: nounwind
declare i32 @__kmpc_global_thread_num(%struct.ident_t*) local_unnamed_addr #3 declare i32 @__kmpc_global_thread_num(%struct.ident_t*) local_unnamed_addr #3
; Function Attrs: alwaysinline ; Function Attrs: alwaysinline
declare void @__kmpc_parallel_51(%struct.ident_t*, i32, i32, i32, i32, i8*, i8*, i8**, i64) local_unnamed_addr #4 declare void @__kmpc_parallel_51(%struct.ident_t*, i32, i32, i32, i32, i8*, i8*, i8**, i64) local_unnamed_addr #4
declare void @__kmpc_target_deinit(%struct.ident_t*, i8, i1) local_unnamed_addr declare void @__kmpc_target_deinit(%struct.ident_t*, i8) local_unnamed_addr
; Function Attrs: convergent ; Function Attrs: convergent
declare double @__nv_sin(double) local_unnamed_addr #5 declare double @__nv_sin(double) local_unnamed_addr #5
attributes #0 = { alwaysinline convergent norecurse nounwind "frame-pointer"="all" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" } attributes #0 = { alwaysinline convergent norecurse nounwind "frame-pointer"="all" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" }
attributes #1 = { alwaysinline mustprogress nofree norecurse nosync nounwind readnone willreturn "frame-pointer"="all" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" } attributes #1 = { alwaysinline mustprogress nofree norecurse nosync nounwind readnone willreturn "frame-pointer"="all" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" }
attributes #2 = { norecurse nounwind "frame-pointer"="all" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" } attributes #2 = { norecurse nounwind "frame-pointer"="all" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" }
attributes #3 = { nounwind } attributes #3 = { nounwind }
▲ Show 20 Lines • Show All 44 Lines • Show Last 20 Lines

llvm/test/Transforms/OpenMP/spmdization_constant_prop.ll

; RUN: opt -S -passes=openmp-opt < %s | FileCheck %s ; RUN: opt -S -passes=openmp-opt < %s | FileCheck %s
; ;
; Verify we change it to SPMD mode but also avoid propagating the old mode (=generic) into the __kmpc_target_init function. ; Verify we change it to SPMD mode but also avoid propagating the old mode (=generic) into the __kmpc_target_init function.
; ;
; CHECK-NOT: store i32 0, ptr addrspace(3) @IsSPMDMode ; CHECK-NOT: store i32 0, ptr addrspace(3) @IsSPMDMode
; CHECK: call i32 @__kmpc_target_init(ptr addrspacecast (ptr addrspace(1) @1 to ptr), i8 2, i1 false, i1 false) ; CHECK: call i32 @__kmpc_target_init(ptr addrspacecast (ptr addrspace(1) @1 to ptr), i8 2, i1 false)
; CHECK-NOT: store i32 0, ptr addrspace(3) @IsSPMDMode ; CHECK-NOT: store i32 0, ptr addrspace(3) @IsSPMDMode
; CHECK: store i32 1, ptr addrspace(3) @IsSPMDMode ; CHECK: store i32 1, ptr addrspace(3) @IsSPMDMode
; CHECK-NOT: store i32 0, ptr addrspace(3) @IsSPMDMode ; CHECK-NOT: store i32 0, ptr addrspace(3) @IsSPMDMode
; ;
target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7" target datalayout = "e-p:64:64-p1:64:64-p2:32:32-p3:32:32-p4:64:64-p5:32:32-p6:32:32-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-v2048:2048-n32:64-S32-A5-G1-ni:7"
target triple = "amdgcn-amd-amdhsa" target triple = "amdgcn-amd-amdhsa"
%struct.ident_t = type { i32, i32, i32, i32, ptr } %struct.ident_t = type { i32, i32, i32, i32, ptr }
Show All 21 Lines
@llvm.used = appending addrspace(1) global [2 x ptr] [ptr addrspacecast (ptr addrspace(3) @IsSPMDMode to ptr), ptr addrspacecast (ptr addrspace(4) @omptarget_device_environment to ptr)], section "llvm.metadata" @llvm.used = appending addrspace(1) global [2 x ptr] [ptr addrspacecast (ptr addrspace(3) @IsSPMDMode to ptr), ptr addrspacecast (ptr addrspace(4) @omptarget_device_environment to ptr)], section "llvm.metadata"
@llvm.compiler.used = appending addrspace(1) global [1 x ptr] [ptr addrspacecast (ptr addrspace(1) @__omp_offloading_20_11e3950_main_l12_exec_mode to ptr)], section "llvm.metadata" @llvm.compiler.used = appending addrspace(1) global [1 x ptr] [ptr addrspacecast (ptr addrspace(1) @__omp_offloading_20_11e3950_main_l12_exec_mode to ptr)], section "llvm.metadata"
; Function Attrs: alwaysinline convergent norecurse nounwind ; Function Attrs: alwaysinline convergent norecurse nounwind
define weak_odr amdgpu_kernel void @__omp_offloading_20_11e3950_main_l12(i64 noundef %nxyz, i64 noundef %ng, ptr noundef nonnull align 8 dereferenceable(8) %aa) local_unnamed_addr #0 { define weak_odr amdgpu_kernel void @__omp_offloading_20_11e3950_main_l12(i64 noundef %nxyz, i64 noundef %ng, ptr noundef nonnull align 8 dereferenceable(8) %aa) local_unnamed_addr #0 {
entry: entry:
%ng1 = alloca i32, align 4 %ng1 = alloca i32, align 4
%captured_vars_addrs = alloca [2 x ptr], align 8, addrspace(5) %captured_vars_addrs = alloca [2 x ptr], align 8, addrspace(5)
%0 = call i32 @__kmpc_target_init(ptr addrspacecast (ptr addrspace(1) @1 to ptr), i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(ptr addrspacecast (ptr addrspace(1) @1 to ptr), i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %common.ret br i1 %exec_user_code, label %user_code.entry, label %common.ret
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%captured_vars_addrs.ascast = addrspacecast ptr addrspace(5) %captured_vars_addrs to ptr %captured_vars_addrs.ascast = addrspacecast ptr addrspace(5) %captured_vars_addrs to ptr
store ptr %ng1, ptr addrspace(5) %captured_vars_addrs, align 8, !tbaa !7 store ptr %ng1, ptr addrspace(5) %captured_vars_addrs, align 8, !tbaa !7
call void @__kmpc_parallel_51(ptr addrspacecast (ptr addrspace(1) @1 to ptr), i32 0, i32 1, i32 -1, i32 -1, ptr nonnull @__omp_outlined__, ptr nonnull @__omp_outlined___wrapper, ptr nonnull %captured_vars_addrs.ascast, i64 2) call void @__kmpc_parallel_51(ptr addrspacecast (ptr addrspace(1) @1 to ptr), i32 0, i32 1, i32 -1, i32 -1, ptr nonnull @__omp_outlined__, ptr nonnull @__omp_outlined___wrapper, ptr nonnull %captured_vars_addrs.ascast, i64 2)
call void @__kmpc_target_deinit(ptr addrspacecast (ptr addrspace(1) @1 to ptr), i8 1, i1 true) call void @__kmpc_target_deinit(ptr addrspacecast (ptr addrspace(1) @1 to ptr), i8 1)
br label %common.ret br label %common.ret
common.ret: ; preds = %user_code.entry, %entry common.ret: ; preds = %user_code.entry, %entry
ret void ret void
} }
; Function Attrs: argmemonly nocallback nofree nosync nounwind willreturn ; Function Attrs: argmemonly nocallback nofree nosync nounwind willreturn
declare void @llvm.lifetime.start.p5(i64 immarg, ptr addrspace(5) nocapture) #1 declare void @llvm.lifetime.start.p5(i64 immarg, ptr addrspace(5) nocapture) #1
Show All 38 Lines
; Function Attrs: convergent mustprogress noinline nounwind willreturn ; Function Attrs: convergent mustprogress noinline nounwind willreturn
define internal fastcc void @_ZN4_OMP11synchronize14threadsAlignedEv() unnamed_addr #8 { define internal fastcc void @_ZN4_OMP11synchronize14threadsAlignedEv() unnamed_addr #8 {
entry: entry:
call void @llvm.amdgcn.s.barrier() #13 call void @llvm.amdgcn.s.barrier() #13
ret void ret void
} }
; Function Attrs: convergent nounwind ; Function Attrs: convergent nounwind
define internal i32 @__kmpc_target_init(ptr nocapture noundef readnone %Ident, i8 noundef signext %Mode, i1 noundef zeroext %UseGenericStateMachine, i1 noundef zeroext %0) local_unnamed_addr #9 { define internal i32 @__kmpc_target_init(ptr nocapture noundef readnone %Ident, i8 noundef signext %Mode, i1 noundef zeroext %UseGenericStateMachine) local_unnamed_addr #9 {
entry: entry:
%0 = and i32 undef, undef
%1 = and i8 %Mode, 2 %1 = and i8 %Mode, 2
%tobool.not = icmp eq i8 %1, 0 %tobool.not = icmp eq i8 %1, 0
br i1 %tobool.not, label %if.else, label %if.then br i1 %tobool.not, label %if.else, label %if.then
if.then: ; preds = %entry if.then: ; preds = %entry
%2 = call i32 @llvm.amdgcn.workitem.id.x() #13, !range !11 %2 = call i32 @llvm.amdgcn.workitem.id.x() #13, !range !11
%3 = call i32 @llvm.amdgcn.workgroup.id.x() #13 %3 = call i32 @llvm.amdgcn.workgroup.id.x() #13
%4 = call align 4 dereferenceable(64) ptr addrspace(4) @llvm.amdgcn.dispatch.ptr() #13 %4 = call align 4 dereferenceable(64) ptr addrspace(4) @llvm.amdgcn.dispatch.ptr() #13
▲ Show 20 Lines • Show All 138 Lines • ▼ Show 20 Linesdo.body.i: ; preds = %if.end10
br label %_ZN14DebugEntryRAIID2Ev.exit250 br label %_ZN14DebugEntryRAIID2Ev.exit250
_ZN14DebugEntryRAIID2Ev.exit250: ; preds = %do.body.i, %if.end10, %_ZN4_OMP7mapping23isInitialThreadInLevel0Eb.exit, %_ZN4_OMP7mapping12getBlockSizeEb.exit.i64 _ZN14DebugEntryRAIID2Ev.exit250: ; preds = %do.body.i, %if.end10, %_ZN4_OMP7mapping23isInitialThreadInLevel0Eb.exit, %_ZN4_OMP7mapping12getBlockSizeEb.exit.i64
%retval.0 = phi i32 [ -1, %_ZN4_OMP7mapping12getBlockSizeEb.exit.i64 ], [ -1, %_ZN4_OMP7mapping23isInitialThreadInLevel0Eb.exit ], [ %25, %do.body.i ], [ %25, %if.end10 ] %retval.0 = phi i32 [ -1, %_ZN4_OMP7mapping12getBlockSizeEb.exit.i64 ], [ -1, %_ZN4_OMP7mapping23isInitialThreadInLevel0Eb.exit ], [ %25, %do.body.i ], [ %25, %if.end10 ]
ret i32 %retval.0 ret i32 %retval.0
} }
; Function Attrs: nounwind ; Function Attrs: nounwind
define internal void @__kmpc_target_deinit(ptr nocapture noundef readnone %Ident, i8 noundef signext %Mode, i1 noundef zeroext %0) local_unnamed_addr #10 { define internal void @__kmpc_target_deinit(ptr nocapture noundef readnone %Ident, i8 noundef signext %Mode) local_unnamed_addr #10 {
ret void ret void
} }
; Function Attrs: convergent nounwind ; Function Attrs: convergent nounwind
declare void @__kmpc_parallel_51(ptr nocapture noundef readnone %ident, i32 noundef %0, i32 noundef %if_expr, i32 noundef %num_threads, i32 noundef %proc_bind, ptr noundef %fn, ptr noundef %wrapper_fn, ptr noundef %args, i64 noundef %nargs) declare void @__kmpc_parallel_51(ptr nocapture noundef readnone %ident, i32 noundef %0, i32 noundef %if_expr, i32 noundef %num_threads, i32 noundef %proc_bind, ptr noundef %fn, ptr noundef %wrapper_fn, ptr noundef %args, i64 noundef %nargs)
; Function Attrs: argmemonly nofree nounwind willreturn ; Function Attrs: argmemonly nofree nounwind willreturn
declare void @llvm.memcpy.p0.p0.i64(ptr noalias nocapture writeonly, ptr noalias nocapture readonly, i64, i1 immarg) #12 declare void @llvm.memcpy.p0.p0.i64(ptr noalias nocapture writeonly, ptr noalias nocapture readonly, i64, i1 immarg) #12
▲ Show 20 Lines • Show All 55 Lines • Show Last 20 Lines

llvm/test/Transforms/OpenMP/spmdization_guarding.ll

Show First 20 Lines • Show All 55 Lines • ▼ Show 20 Lines
; CHECK-DISABLED: @[[__OMP_OUTLINED__1_WRAPPER_ID:[a-zA-Z0-9_$"\\.-]+]] = private constant i8 undef ; CHECK-DISABLED: @[[__OMP_OUTLINED__1_WRAPPER_ID:[a-zA-Z0-9_$"\\.-]+]] = private constant i8 undef
;. ;.
define weak void @__omp_offloading_2a_fbfa7a_sequential_loop_l6(i32* %x, i64 %N) #0 { define weak void @__omp_offloading_2a_fbfa7a_sequential_loop_l6(i32* %x, i64 %N) #0 {
; CHECK-LABEL: define {{[^@]+}}@__omp_offloading_2a_fbfa7a_sequential_loop_l6 ; CHECK-LABEL: define {{[^@]+}}@__omp_offloading_2a_fbfa7a_sequential_loop_l6
; CHECK-SAME: (i32* [[X:%.*]], i64 [[N:%.*]]) #[[ATTR0:[0-9]+]] { ; CHECK-SAME: (i32* [[X:%.*]], i64 [[N:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: [[HEAP2STACK_H2S:%.*]] = alloca i8, i64 8, align 8 ; CHECK-NEXT: [[HEAP2STACK_H2S:%.*]] = alloca i8, i64 8, align 8
; CHECK-NEXT: [[N_ADDR_SROA_0_0_EXTRACT_TRUNC:%.*]] = trunc i64 [[N]] to i32 ; CHECK-NEXT: [[N_ADDR_SROA_0_0_EXTRACT_TRUNC:%.*]] = trunc i64 [[N]] to i32
; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @[[GLOB1]], i8 2, i1 false, i1 false) #[[ATTR6:[0-9]+]] ; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @[[GLOB1]], i8 2, i1 false) #[[ATTR6:[0-9]+]]
; CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; 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-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; CHECK: user_code.entry: ; CHECK: user_code.entry:
; CHECK-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @[[GLOB1]]) #[[ATTR6]] ; CHECK-NEXT: [[TMP1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @[[GLOB1]]) #[[ATTR6]]
; CHECK-NEXT: [[ARRAYIDX1_I:%.*]] = getelementptr inbounds i32, i32* [[X]], i64 1 ; CHECK-NEXT: [[ARRAYIDX1_I:%.*]] = getelementptr inbounds i32, i32* [[X]], i64 1
; CHECK-NEXT: [[SEXT:%.*]] = shl i64 [[N]], 32 ; CHECK-NEXT: [[SEXT:%.*]] = shl i64 [[N]], 32
; CHECK-NEXT: [[IDXPROM_I:%.*]] = ashr exact i64 [[SEXT]], 32 ; CHECK-NEXT: [[IDXPROM_I:%.*]] = ashr exact i64 [[SEXT]], 32
; CHECK-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds i32, i32* [[X]], i64 [[IDXPROM_I]] ; CHECK-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds i32, i32* [[X]], i64 [[IDXPROM_I]]
▲ Show 20 Lines • Show All 91 Lines • ▼ Show 20 Lines
; CHECK-NEXT: br label [[REGION_BARRIER17]] ; CHECK-NEXT: br label [[REGION_BARRIER17]]
; CHECK: region.barrier17: ; CHECK: region.barrier17:
; CHECK-NEXT: call void @__kmpc_barrier_simple_spmd(%struct.ident_t* @[[GLOB2]], i32 [[TMP10]]) ; CHECK-NEXT: call void @__kmpc_barrier_simple_spmd(%struct.ident_t* @[[GLOB2]], i32 [[TMP10]])
; CHECK-NEXT: br label [[REGION_EXIT18:%.*]] ; CHECK-NEXT: br label [[REGION_EXIT18:%.*]]
; CHECK: region.exit18: ; CHECK: region.exit18:
; CHECK-NEXT: [[CALL14_I:%.*]] = call i32 @no_openmp(i32* nonnull [[X]]) #[[ATTR10]], !noalias !8 ; CHECK-NEXT: [[CALL14_I:%.*]] = call i32 @no_openmp(i32* nonnull [[X]]) #[[ATTR10]], !noalias !8
; CHECK-NEXT: [[CALL15_I:%.*]] = call i32 @no_openmp(i32* nonnull [[X]]) #[[ATTR10]], !noalias !8 ; CHECK-NEXT: [[CALL15_I:%.*]] = call i32 @no_openmp(i32* nonnull [[X]]) #[[ATTR10]], !noalias !8
; CHECK-NEXT: [[CALL16_I:%.*]] = call i32 @no_openmp(i32* nonnull [[X]]) #[[ATTR10]], !noalias !8 ; CHECK-NEXT: [[CALL16_I:%.*]] = call i32 @no_openmp(i32* nonnull [[X]]) #[[ATTR10]], !noalias !8
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* nonnull @[[GLOB1]], i8 2, i1 false) #[[ATTR6]] ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* nonnull @[[GLOB1]], i8 2) #[[ATTR6]]
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; CHECK: worker.exit: ; CHECK: worker.exit:
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
; CHECK-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_2a_fbfa7a_sequential_loop_l6 ; CHECK-DISABLED-LABEL: define {{[^@]+}}@__omp_offloading_2a_fbfa7a_sequential_loop_l6
; CHECK-DISABLED-SAME: (i32* [[X:%.*]], i64 [[N:%.*]]) #[[ATTR0:[0-9]+]] { ; CHECK-DISABLED-SAME: (i32* [[X:%.*]], i64 [[N:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-DISABLED-NEXT: entry: ; CHECK-DISABLED-NEXT: entry:
; CHECK-DISABLED-NEXT: [[HEAP2STACK_H2S:%.*]] = alloca i8, i64 8, align 8 ; CHECK-DISABLED-NEXT: [[HEAP2STACK_H2S:%.*]] = alloca i8, i64 8, align 8
; CHECK-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8 ; CHECK-DISABLED-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8
; CHECK-DISABLED-NEXT: [[N_ADDR_SROA_0_0_EXTRACT_TRUNC:%.*]] = trunc i64 [[N]] to i32 ; CHECK-DISABLED-NEXT: [[N_ADDR_SROA_0_0_EXTRACT_TRUNC:%.*]] = trunc i64 [[N]] to i32
; CHECK-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @[[GLOB1]], i8 1, i1 false, i1 true) #[[ATTR6:[0-9]+]] ; CHECK-DISABLED-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @[[GLOB1]], i8 1, i1 false) #[[ATTR6:[0-9]+]]
; CHECK-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; CHECK-DISABLED-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; CHECK-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; CHECK-DISABLED-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; CHECK-DISABLED: is_worker_check: ; CHECK-DISABLED: is_worker_check:
; CHECK-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; CHECK-DISABLED-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; CHECK-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; CHECK-DISABLED-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; CHECK-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; CHECK-DISABLED-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; CHECK-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; CHECK-DISABLED-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; CHECK-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; CHECK-DISABLED-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
▲ Show 20 Lines • Show All 59 Lines • ▼ Show 20 Lines
; CHECK-DISABLED-NEXT: store i32 [[CALL8_I]], i32* [[ARRAYIDX10_I]], align 4, !noalias !8 ; CHECK-DISABLED-NEXT: store i32 [[CALL8_I]], i32* [[ARRAYIDX10_I]], align 4, !noalias !8
; CHECK-DISABLED-NEXT: [[CALL11_I:%.*]] = call i32 @no_openmp(i32* nonnull [[X]]) #[[ATTR10]], !noalias !8 ; CHECK-DISABLED-NEXT: [[CALL11_I:%.*]] = call i32 @no_openmp(i32* nonnull [[X]]) #[[ATTR10]], !noalias !8
; CHECK-DISABLED-NEXT: [[IDXPROM12_I:%.*]] = sext i32 [[CALL11_I]] to i64 ; CHECK-DISABLED-NEXT: [[IDXPROM12_I:%.*]] = sext i32 [[CALL11_I]] to i64
; CHECK-DISABLED-NEXT: [[ARRAYIDX13_I:%.*]] = getelementptr inbounds i32, i32* [[X]], i64 [[IDXPROM12_I]] ; CHECK-DISABLED-NEXT: [[ARRAYIDX13_I:%.*]] = getelementptr inbounds i32, i32* [[X]], i64 [[IDXPROM12_I]]
; CHECK-DISABLED-NEXT: store i32 [[CALL11_I]], i32* [[ARRAYIDX13_I]], align 4, !noalias !8 ; CHECK-DISABLED-NEXT: store i32 [[CALL11_I]], i32* [[ARRAYIDX13_I]], align 4, !noalias !8
; CHECK-DISABLED-NEXT: [[CALL14_I:%.*]] = call i32 @no_openmp(i32* nonnull [[X]]) #[[ATTR10]], !noalias !8 ; CHECK-DISABLED-NEXT: [[CALL14_I:%.*]] = call i32 @no_openmp(i32* nonnull [[X]]) #[[ATTR10]], !noalias !8
; CHECK-DISABLED-NEXT: [[CALL15_I:%.*]] = call i32 @no_openmp(i32* nonnull [[X]]) #[[ATTR10]], !noalias !8 ; CHECK-DISABLED-NEXT: [[CALL15_I:%.*]] = call i32 @no_openmp(i32* nonnull [[X]]) #[[ATTR10]], !noalias !8
; CHECK-DISABLED-NEXT: [[CALL16_I:%.*]] = call i32 @no_openmp(i32* nonnull [[X]]) #[[ATTR10]], !noalias !8 ; CHECK-DISABLED-NEXT: [[CALL16_I:%.*]] = call i32 @no_openmp(i32* nonnull [[X]]) #[[ATTR10]], !noalias !8
; CHECK-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* nonnull @[[GLOB1]], i8 1, i1 true) #[[ATTR6]] ; CHECK-DISABLED-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* nonnull @[[GLOB1]], i8 1) #[[ATTR6]]
; CHECK-DISABLED-NEXT: ret void ; CHECK-DISABLED-NEXT: ret void
; CHECK-DISABLED: worker.exit: ; CHECK-DISABLED: worker.exit:
; CHECK-DISABLED-NEXT: ret void ; CHECK-DISABLED-NEXT: ret void
; ;
entry: entry:
%N.addr.sroa.0.0.extract.trunc = trunc i64 %N to i32 %N.addr.sroa.0.0.extract.trunc = trunc i64 %N to i32
%0 = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @1, i8 1, i1 true, i1 true) #3 %0 = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @1, i8 1, i1 true) #3
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1) %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @1)
store i32 0, i32* %x, align 4, !noalias !8 store i32 0, i32* %x, align 4, !noalias !8
%arrayidx1.i = getelementptr inbounds i32, i32* %x, i64 1 %arrayidx1.i = getelementptr inbounds i32, i32* %x, i64 1
store i32 1, i32* %arrayidx1.i, align 4, !noalias !8 store i32 1, i32* %arrayidx1.i, align 4, !noalias !8
Show All 32 Lines__omp_outlined__.exit: ; preds = %for.cond.i
store i32 %call8.i, i32* %arrayidx10.i, align 4, !noalias !8 store i32 %call8.i, i32* %arrayidx10.i, align 4, !noalias !8
%call11.i = call i32 @no_openmp(i32* nonnull %x) #5, !noalias !8 %call11.i = call i32 @no_openmp(i32* nonnull %x) #5, !noalias !8
%idxprom12.i = sext i32 %call11.i to i64 %idxprom12.i = sext i32 %call11.i to i64
%arrayidx13.i = getelementptr inbounds i32, i32* %x, i64 %idxprom12.i %arrayidx13.i = getelementptr inbounds i32, i32* %x, i64 %idxprom12.i
store i32 %call11.i, i32* %arrayidx13.i, align 4, !noalias !8 store i32 %call11.i, i32* %arrayidx13.i, align 4, !noalias !8
%call14.i = call i32 @no_openmp(i32* nonnull %x) #5, !noalias !8 %call14.i = call i32 @no_openmp(i32* nonnull %x) #5, !noalias !8
%call15.i = call i32 @no_openmp(i32* nonnull %x) #5, !noalias !8 %call15.i = call i32 @no_openmp(i32* nonnull %x) #5, !noalias !8
%call16.i = call i32 @no_openmp(i32* nonnull %x) #5, !noalias !8 %call16.i = call i32 @no_openmp(i32* nonnull %x) #5, !noalias !8
call void @__kmpc_target_deinit(%struct.ident_t* nonnull @1, i8 1, i1 true) #3 call void @__kmpc_target_deinit(%struct.ident_t* nonnull @1, i8 1) #3
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
define internal void @__omp_outlined__1(i32* noalias %.global_tid., i32* noalias %.bound_tid.) { define internal void @__omp_outlined__1(i32* noalias %.global_tid., i32* noalias %.bound_tid.) {
; CHECK-LABEL: define {{[^@]+}}@__omp_outlined__1 ; CHECK-LABEL: define {{[^@]+}}@__omp_outlined__1
Show All 16 Lines
; CHECK-DISABLED-NEXT: ret void ; CHECK-DISABLED-NEXT: ret void
; ;
ret void ret void
} }
declare void @__kmpc_parallel_51(%struct.ident_t*, i32, i32, i32, i32, i8*, i8*, i8**, i64) declare void @__kmpc_parallel_51(%struct.ident_t*, i32, i32, i32, i32, i8*, i8*, i8**, i64)
; Make it a weak definition so we will apply custom state machine rewriting but can't use the body in the reasoning. ; Make it a weak definition so we will apply custom state machine rewriting but can't use the body in the reasoning.
define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1, i1) { define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1) {
; CHECK-LABEL: define {{[^@]+}}@__kmpc_target_init
; CHECK-SAME: (%struct.ident_t* [[TMP0:%.*]], i8 [[TMP1:%.*]], i1 [[TMP2:%.*]]) {
; CHECK-NEXT: ret i32 0
;
; CHECK-DISABLED-LABEL: define {{[^@]+}}@__kmpc_target_init
; CHECK-DISABLED-SAME: (%struct.ident_t* [[TMP0:%.*]], i8 [[TMP1:%.*]], i1 [[TMP2:%.*]]) {
; CHECK-DISABLED-NEXT: ret i32 0
;
ret i32 0 ret i32 0
} }
; Function Attrs: convergent ; Function Attrs: convergent
declare i32 @no_openmp(i32*) #1 declare i32 @no_openmp(i32*) #1
; Function Attrs: convergent nounwind readonly willreturn ; Function Attrs: convergent nounwind readonly willreturn
declare void @pure() #2 declare void @pure() #2
declare i8* @__kmpc_alloc_shared(i64) declare i8* @__kmpc_alloc_shared(i64)
declare void @__kmpc_free_shared(i8* nocapture, i64) declare void @__kmpc_free_shared(i8* nocapture, i64)
declare void @usei8ptr(i8*) #1 declare void @usei8ptr(i8*) #1
; Function Attrs: nounwind ; Function Attrs: nounwind
declare i32 @__kmpc_global_thread_num(%struct.ident_t*) #3 declare i32 @__kmpc_global_thread_num(%struct.ident_t*) #3
declare void @__kmpc_target_deinit(%struct.ident_t*, i8, i1) declare void @__kmpc_target_deinit(%struct.ident_t*, i8)
; Function Attrs: inaccessiblememonly nofree nosync nounwind willreturn ; Function Attrs: inaccessiblememonly nofree nosync nounwind willreturn
declare void @llvm.experimental.noalias.scope.decl(metadata) #4 declare void @llvm.experimental.noalias.scope.decl(metadata) #4
attributes #0 = { convergent norecurse nounwind "frame-pointer"="all" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="sm_53" "target-features"="+ptx32,+sm_53" } attributes #0 = { convergent norecurse nounwind "frame-pointer"="all" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="sm_53" "target-features"="+ptx32,+sm_53" }
attributes #1 = { convergent "frame-pointer"="all" "llvm.assume"="omp_no_openmp,ompx_spmd_amenable" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="sm_53" "target-features"="+ptx32,+sm_53" } attributes #1 = { convergent "frame-pointer"="all" "llvm.assume"="omp_no_openmp,ompx_spmd_amenable" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="sm_53" "target-features"="+ptx32,+sm_53" }
attributes #2 = { convergent nounwind readonly willreturn "frame-pointer"="all" "llvm.assume"="ompx_spmd_amenable" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="sm_53" "target-features"="+ptx32,+sm_53" } attributes #2 = { convergent nounwind readonly willreturn "frame-pointer"="all" "llvm.assume"="ompx_spmd_amenable" "no-trapping-math"="true" "stack-protector-buffer-size"="8" "target-cpu"="sm_53" "target-features"="+ptx32,+sm_53" }
attributes #3 = { nounwind } attributes #3 = { nounwind }
▲ Show 20 Lines • Show All 74 Lines • Show Last 20 Lines

llvm/test/Transforms/OpenMP/spmdization_guarding_two_reaching_kernels.ll

Show First 20 Lines • Show All 56 Lines • ▼ Show 20 Lines
; CHECK-DISABLE-SPMDIZATION: @[[G:[a-zA-Z0-9_$"\\.-]+]] = external global i32, align 4 ; CHECK-DISABLE-SPMDIZATION: @[[G:[a-zA-Z0-9_$"\\.-]+]] = external global i32, align 4
; CHECK-DISABLE-SPMDIZATION: @[[LLVM_COMPILER_USED:[a-zA-Z0-9_$"\\.-]+]] = appending global [2 x i8*] [i8* @__omp_offloading_2b_10393b5_spmd_l12_exec_mode, i8* @__omp_offloading_2b_10393b5_generic_l20_exec_mode], section "llvm.metadata" ; CHECK-DISABLE-SPMDIZATION: @[[LLVM_COMPILER_USED:[a-zA-Z0-9_$"\\.-]+]] = appending global [2 x i8*] [i8* @__omp_offloading_2b_10393b5_spmd_l12_exec_mode, i8* @__omp_offloading_2b_10393b5_generic_l20_exec_mode], section "llvm.metadata"
; CHECK-DISABLE-SPMDIZATION: @[[__OMP_OUTLINED___WRAPPER_ID:[a-zA-Z0-9_$"\\.-]+]] = private constant i8 undef ; CHECK-DISABLE-SPMDIZATION: @[[__OMP_OUTLINED___WRAPPER_ID:[a-zA-Z0-9_$"\\.-]+]] = private constant i8 undef
;. ;.
define weak void @__omp_offloading_2b_10393b5_spmd_l12() #0 { define weak void @__omp_offloading_2b_10393b5_spmd_l12() #0 {
; CHECK-LABEL: define {{[^@]+}}@__omp_offloading_2b_10393b5_spmd_l12 ; CHECK-LABEL: define {{[^@]+}}@__omp_offloading_2b_10393b5_spmd_l12
; CHECK-SAME: () #[[ATTR0:[0-9]+]] { ; CHECK-SAME: () #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false, i1 false) ; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false)
; CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; 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-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; CHECK: user_code.entry: ; CHECK: user_code.entry:
; CHECK-NEXT: call void @spmd_helper() #[[ATTR5:[0-9]+]] ; CHECK-NEXT: call void @spmd_helper() #[[ATTR5:[0-9]+]]
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 false) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; CHECK: worker.exit: ; CHECK: worker.exit:
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
; CHECK-DISABLE-SPMDIZATION-LABEL: define {{[^@]+}}@__omp_offloading_2b_10393b5_spmd_l12 ; CHECK-DISABLE-SPMDIZATION-LABEL: define {{[^@]+}}@__omp_offloading_2b_10393b5_spmd_l12
; CHECK-DISABLE-SPMDIZATION-SAME: () #[[ATTR0:[0-9]+]] { ; CHECK-DISABLE-SPMDIZATION-SAME: () #[[ATTR0:[0-9]+]] {
; CHECK-DISABLE-SPMDIZATION-NEXT: entry: ; CHECK-DISABLE-SPMDIZATION-NEXT: entry:
; CHECK-DISABLE-SPMDIZATION-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8 ; CHECK-DISABLE-SPMDIZATION-NEXT: [[WORKER_WORK_FN_ADDR:%.*]] = alloca i8*, align 8
; CHECK-DISABLE-SPMDIZATION-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; CHECK-DISABLE-SPMDIZATION-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; CHECK-DISABLE-SPMDIZATION-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1 ; CHECK-DISABLE-SPMDIZATION-NEXT: [[THREAD_IS_WORKER:%.*]] = icmp ne i32 [[TMP0]], -1
; CHECK-DISABLE-SPMDIZATION-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]] ; CHECK-DISABLE-SPMDIZATION-NEXT: br i1 [[THREAD_IS_WORKER]], label [[IS_WORKER_CHECK:%.*]], label [[THREAD_USER_CODE_CHECK:%.*]]
; CHECK-DISABLE-SPMDIZATION: is_worker_check: ; CHECK-DISABLE-SPMDIZATION: is_worker_check:
; CHECK-DISABLE-SPMDIZATION-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block() ; CHECK-DISABLE-SPMDIZATION-NEXT: [[BLOCK_HW_SIZE:%.*]] = call i32 @__kmpc_get_hardware_num_threads_in_block()
; CHECK-DISABLE-SPMDIZATION-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size() ; CHECK-DISABLE-SPMDIZATION-NEXT: [[WARP_SIZE:%.*]] = call i32 @__kmpc_get_warp_size()
; CHECK-DISABLE-SPMDIZATION-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]] ; CHECK-DISABLE-SPMDIZATION-NEXT: [[BLOCK_SIZE:%.*]] = sub i32 [[BLOCK_HW_SIZE]], [[WARP_SIZE]]
; CHECK-DISABLE-SPMDIZATION-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]] ; CHECK-DISABLE-SPMDIZATION-NEXT: [[THREAD_IS_MAIN_OR_WORKER:%.*]] = icmp slt i32 [[TMP0]], [[BLOCK_SIZE]]
; CHECK-DISABLE-SPMDIZATION-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]] ; CHECK-DISABLE-SPMDIZATION-NEXT: br i1 [[THREAD_IS_MAIN_OR_WORKER]], label [[WORKER_STATE_MACHINE_BEGIN:%.*]], label [[WORKER_STATE_MACHINE_FINISHED:%.*]]
Show All 21 Lines
; CHECK-DISABLE-SPMDIZATION: worker_state_machine.done.barrier: ; CHECK-DISABLE-SPMDIZATION: worker_state_machine.done.barrier:
; CHECK-DISABLE-SPMDIZATION-NEXT: call void @__kmpc_barrier_simple_generic(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]]) ; CHECK-DISABLE-SPMDIZATION-NEXT: call void @__kmpc_barrier_simple_generic(%struct.ident_t* @[[GLOB1]], i32 [[TMP0]])
; CHECK-DISABLE-SPMDIZATION-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]] ; CHECK-DISABLE-SPMDIZATION-NEXT: br label [[WORKER_STATE_MACHINE_BEGIN]]
; CHECK-DISABLE-SPMDIZATION: thread.user_code.check: ; CHECK-DISABLE-SPMDIZATION: thread.user_code.check:
; CHECK-DISABLE-SPMDIZATION-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; CHECK-DISABLE-SPMDIZATION-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; CHECK-DISABLE-SPMDIZATION-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; CHECK-DISABLE-SPMDIZATION-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; CHECK-DISABLE-SPMDIZATION: user_code.entry: ; CHECK-DISABLE-SPMDIZATION: user_code.entry:
; CHECK-DISABLE-SPMDIZATION-NEXT: call void @spmd_helper() #[[ATTR5:[0-9]+]] ; CHECK-DISABLE-SPMDIZATION-NEXT: call void @spmd_helper() #[[ATTR5:[0-9]+]]
; CHECK-DISABLE-SPMDIZATION-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; CHECK-DISABLE-SPMDIZATION-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; CHECK-DISABLE-SPMDIZATION-NEXT: ret void ; CHECK-DISABLE-SPMDIZATION-NEXT: ret void
; CHECK-DISABLE-SPMDIZATION: worker.exit: ; CHECK-DISABLE-SPMDIZATION: worker.exit:
; CHECK-DISABLE-SPMDIZATION-NEXT: ret void ; CHECK-DISABLE-SPMDIZATION-NEXT: ret void
; ;
entry: entry:
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
call void @spmd_helper() #5 call void @spmd_helper() #5
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
; Make it a weak definition so we will apply custom state machine rewriting but can't use the body in the reasoning. ; Make it a weak definition so we will apply custom state machine rewriting but can't use the body in the reasoning.
define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1, i1) { define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1) {
; CHECK-LABEL: define {{[^@]+}}@__kmpc_target_init
; CHECK-SAME: (%struct.ident_t* [[TMP0:%.*]], i8 [[TMP1:%.*]], i1 [[TMP2:%.*]]) {
; CHECK-NEXT: ret i32 0
;
; CHECK-DISABLE-SPMDIZATION-LABEL: define {{[^@]+}}@__kmpc_target_init
; CHECK-DISABLE-SPMDIZATION-SAME: (%struct.ident_t* [[TMP0:%.*]], i8 [[TMP1:%.*]], i1 [[TMP2:%.*]]) {
; CHECK-DISABLE-SPMDIZATION-NEXT: ret i32 0
;
ret i32 0 ret i32 0
} }
declare void @__kmpc_target_deinit(%struct.ident_t*, i8, i1) declare void @__kmpc_target_deinit(%struct.ident_t*, i8)
; Function Attrs: convergent noinline norecurse nounwind ; Function Attrs: convergent noinline norecurse nounwind
define weak void @__omp_offloading_2b_10393b5_generic_l20() #0 { define weak void @__omp_offloading_2b_10393b5_generic_l20() #0 {
; CHECK-LABEL: define {{[^@]+}}@__omp_offloading_2b_10393b5_generic_l20 ; CHECK-LABEL: define {{[^@]+}}@__omp_offloading_2b_10393b5_generic_l20
; CHECK-SAME: () #[[ATTR0]] { ; CHECK-SAME: () #[[ATTR0]] {
; CHECK-NEXT: entry: ; CHECK-NEXT: entry:
; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false, i1 false) ; CHECK-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 2, i1 false)
; CHECK-NEXT: [[THREAD_ID_IN_BLOCK:%.*]] = call i32 @__kmpc_get_hardware_thread_id_in_block() ; CHECK-NEXT: [[THREAD_ID_IN_BLOCK:%.*]] = call i32 @__kmpc_get_hardware_thread_id_in_block()
; CHECK-NEXT: [[THREAD_IS_MAIN:%.*]] = icmp ne i32 [[THREAD_ID_IN_BLOCK]], 0 ; CHECK-NEXT: [[THREAD_IS_MAIN:%.*]] = icmp ne i32 [[THREAD_ID_IN_BLOCK]], 0
; CHECK-NEXT: br i1 [[THREAD_IS_MAIN]], label [[EXIT_THREADS:%.*]], label [[MAIN_THREAD_USER_CODE:%.*]] ; CHECK-NEXT: br i1 [[THREAD_IS_MAIN]], label [[EXIT_THREADS:%.*]], label [[MAIN_THREAD_USER_CODE:%.*]]
; CHECK: exit.threads: ; CHECK: exit.threads:
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; CHECK: main.thread.user_code: ; CHECK: main.thread.user_code:
; CHECK-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; 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-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; CHECK: user_code.entry: ; CHECK: user_code.entry:
; CHECK-NEXT: call void @generic_helper() #[[ATTR5]] ; CHECK-NEXT: call void @generic_helper() #[[ATTR5]]
; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2, i1 false) ; CHECK-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 2)
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; CHECK: worker.exit: ; CHECK: worker.exit:
; CHECK-NEXT: ret void ; CHECK-NEXT: ret void
; ;
; CHECK-DISABLE-SPMDIZATION-LABEL: define {{[^@]+}}@__omp_offloading_2b_10393b5_generic_l20 ; CHECK-DISABLE-SPMDIZATION-LABEL: define {{[^@]+}}@__omp_offloading_2b_10393b5_generic_l20
; CHECK-DISABLE-SPMDIZATION-SAME: () #[[ATTR0]] { ; CHECK-DISABLE-SPMDIZATION-SAME: () #[[ATTR0]] {
; CHECK-DISABLE-SPMDIZATION-NEXT: entry: ; CHECK-DISABLE-SPMDIZATION-NEXT: entry:
; CHECK-DISABLE-SPMDIZATION-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false, i1 true) ; CHECK-DISABLE-SPMDIZATION-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_target_init(%struct.ident_t* @[[GLOB1]], i8 1, i1 false)
; CHECK-DISABLE-SPMDIZATION-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1 ; CHECK-DISABLE-SPMDIZATION-NEXT: [[EXEC_USER_CODE:%.*]] = icmp eq i32 [[TMP0]], -1
; CHECK-DISABLE-SPMDIZATION-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]] ; CHECK-DISABLE-SPMDIZATION-NEXT: br i1 [[EXEC_USER_CODE]], label [[USER_CODE_ENTRY:%.*]], label [[WORKER_EXIT:%.*]]
; CHECK-DISABLE-SPMDIZATION: user_code.entry: ; CHECK-DISABLE-SPMDIZATION: user_code.entry:
; CHECK-DISABLE-SPMDIZATION-NEXT: call void @generic_helper() #[[ATTR5]] ; CHECK-DISABLE-SPMDIZATION-NEXT: call void @generic_helper() #[[ATTR5]]
; CHECK-DISABLE-SPMDIZATION-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1, i1 true) ; CHECK-DISABLE-SPMDIZATION-NEXT: call void @__kmpc_target_deinit(%struct.ident_t* @[[GLOB1]], i8 1)
; CHECK-DISABLE-SPMDIZATION-NEXT: ret void ; CHECK-DISABLE-SPMDIZATION-NEXT: ret void
; CHECK-DISABLE-SPMDIZATION: worker.exit: ; CHECK-DISABLE-SPMDIZATION: worker.exit:
; CHECK-DISABLE-SPMDIZATION-NEXT: ret void ; CHECK-DISABLE-SPMDIZATION-NEXT: ret void
; ;
entry: entry:
%0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true, i1 true) %0 = call i32 @__kmpc_target_init(%struct.ident_t* @1, i8 1, i1 true)
%exec_user_code = icmp eq i32 %0, -1 %exec_user_code = icmp eq i32 %0, -1
br i1 %exec_user_code, label %user_code.entry, label %worker.exit br i1 %exec_user_code, label %user_code.entry, label %worker.exit
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
call void @generic_helper() #5 call void @generic_helper() #5
call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1, i1 true) call void @__kmpc_target_deinit(%struct.ident_t* @1, i8 1)
ret void ret void
worker.exit: ; preds = %entry worker.exit: ; preds = %entry
ret void ret void
} }
; Function Attrs: convergent noinline nounwind ; Function Attrs: convergent noinline nounwind
define internal void @spmd_helper() #1 { define internal void @spmd_helper() #1 {
▲ Show 20 Lines • Show All 215 Lines • Show Last 20 Lines

llvm/test/Transforms/OpenMP/spmdization_remarks.ll

Show First 20 Lines • Show All 56 Lines • ▼ Show 20 Lines
@13 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 2, i32 0, i8* getelementptr inbounds ([63 x i8], [63 x i8]* @12, i32 0, i32 0) }, align 8 @13 = private unnamed_addr constant %struct.ident_t { i32 0, i32 2, i32 2, i32 0, i8* getelementptr inbounds ([63 x i8], [63 x i8]* @12, i32 0, i32 0) }, align 8
@G = external global i32 @G = external global i32
@llvm.compiler.used = appending global [2 x i8*] [i8* @__omp_offloading_2a_d80d3d_test_fallback_l11_exec_mode, i8* @__omp_offloading_2a_d80d3d_test_no_fallback_l20_exec_mode], section "llvm.metadata" @llvm.compiler.used = appending global [2 x i8*] [i8* @__omp_offloading_2a_d80d3d_test_fallback_l11_exec_mode, i8* @__omp_offloading_2a_d80d3d_test_no_fallback_l20_exec_mode], section "llvm.metadata"
; Function Attrs: convergent norecurse nounwind ; Function Attrs: convergent norecurse nounwind
define weak void @__omp_offloading_2a_d80d3d_test_fallback_l11() local_unnamed_addr #0 !dbg !15 { define weak void @__omp_offloading_2a_d80d3d_test_fallback_l11() local_unnamed_addr #0 !dbg !15 {
entry: entry:
%captured_vars_addrs.i.i = alloca [0 x i8*], align 8 %captured_vars_addrs.i.i = alloca [0 x i8*], align 8
%0 = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @1, i8 1, i1 true, i1 true) #3, !dbg !18 %0 = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @1, i8 1, i1 true) #3, !dbg !18
%exec_user_code = icmp eq i32 %0, -1, !dbg !18 %exec_user_code = icmp eq i32 %0, -1, !dbg !18
br i1 %exec_user_code, label %user_code.entry, label %common.ret, !dbg !18 br i1 %exec_user_code, label %user_code.entry, label %common.ret, !dbg !18
common.ret: ; preds = %entry, %user_code.entry common.ret: ; preds = %entry, %user_code.entry
ret void, !dbg !19 ret void, !dbg !19
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @3) #3 %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @3) #3
call void @unknown() #6, !dbg !20 call void @unknown() #6, !dbg !20
%2 = bitcast [0 x i8*]* %captured_vars_addrs.i.i to i8* %2 = bitcast [0 x i8*]* %captured_vars_addrs.i.i to i8*
call void @llvm.lifetime.start.p0i8(i64 0, i8* nonnull %2) #3 call void @llvm.lifetime.start.p0i8(i64 0, i8* nonnull %2) #3
%3 = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @13) #3 %3 = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @13) #3
%4 = getelementptr inbounds [0 x i8*], [0 x i8*]* %captured_vars_addrs.i.i, i64 0, i64 0, !dbg !23 %4 = getelementptr inbounds [0 x i8*], [0 x i8*]* %captured_vars_addrs.i.i, i64 0, i64 0, !dbg !23
call void @__kmpc_parallel_51(%struct.ident_t* noundef nonnull @13, i32 %3, i32 noundef 1, i32 noundef -1, i32 noundef -1, i8* noundef bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* noundef bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** noundef nonnull %4, i64 noundef 0) #3, !dbg !23 call void @__kmpc_parallel_51(%struct.ident_t* noundef nonnull @13, i32 %3, i32 noundef 1, i32 noundef -1, i32 noundef -1, i8* noundef bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* noundef bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** noundef nonnull %4, i64 noundef 0) #3, !dbg !23
call void @llvm.lifetime.end.p0i8(i64 0, i8* nonnull %2) #3, !dbg !26 call void @llvm.lifetime.end.p0i8(i64 0, i8* nonnull %2) #3, !dbg !26
call void @unknown() #6, !dbg !27 call void @unknown() #6, !dbg !27
call void @__kmpc_target_deinit(%struct.ident_t* nonnull @5, i8 1, i1 true) #3, !dbg !28 call void @__kmpc_target_deinit(%struct.ident_t* nonnull @5, i8 1) #3, !dbg !28
br label %common.ret br label %common.ret
} }
define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1, i1) { define weak i32 @__kmpc_target_init(%struct.ident_t*, i8, i1) {
ret i32 0 ret i32 0
} }
; Function Attrs: convergent ; Function Attrs: convergent
declare void @unknown() local_unnamed_addr #1 declare void @unknown() local_unnamed_addr #1
; Function Attrs: nounwind ; Function Attrs: nounwind
define hidden void @known() local_unnamed_addr #2 !dbg !29 { define hidden void @known() local_unnamed_addr #2 !dbg !29 {
entry: entry:
%captured_vars_addrs = alloca [0 x i8*], align 8 %captured_vars_addrs = alloca [0 x i8*], align 8
%0 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @13) %0 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @13)
%1 = getelementptr inbounds [0 x i8*], [0 x i8*]* %captured_vars_addrs, i64 0, i64 0, !dbg !30 %1 = getelementptr inbounds [0 x i8*], [0 x i8*]* %captured_vars_addrs, i64 0, i64 0, !dbg !30
call void @__kmpc_parallel_51(%struct.ident_t* nonnull @13, i32 %0, i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** nonnull %1, i64 0) #3, !dbg !30 call void @__kmpc_parallel_51(%struct.ident_t* nonnull @13, i32 %0, i32 1, i32 -1, i32 -1, i8* bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** nonnull %1, i64 0) #3, !dbg !30
ret void, !dbg !31 ret void, !dbg !31
} }
; Function Attrs: nounwind ; Function Attrs: nounwind
declare i32 @__kmpc_global_thread_num(%struct.ident_t*) local_unnamed_addr #3 declare i32 @__kmpc_global_thread_num(%struct.ident_t*) local_unnamed_addr #3
declare void @__kmpc_target_deinit(%struct.ident_t*, i8, i1) local_unnamed_addr declare void @__kmpc_target_deinit(%struct.ident_t*, i8) local_unnamed_addr
; Function Attrs: norecurse nounwind ; Function Attrs: norecurse nounwind
define weak void @__omp_offloading_2a_d80d3d_test_no_fallback_l20() local_unnamed_addr #4 !dbg !32 { define weak void @__omp_offloading_2a_d80d3d_test_no_fallback_l20() local_unnamed_addr #4 !dbg !32 {
entry: entry:
%captured_vars_addrs.i2.i = alloca [0 x i8*], align 8 %captured_vars_addrs.i2.i = alloca [0 x i8*], align 8
%0 = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @7, i8 1, i1 true, i1 true) #3, !dbg !33 %0 = call i32 @__kmpc_target_init(%struct.ident_t* nonnull @7, i8 1, i1 true) #3, !dbg !33
%exec_user_code = icmp eq i32 %0, -1, !dbg !33 %exec_user_code = icmp eq i32 %0, -1, !dbg !33
br i1 %exec_user_code, label %user_code.entry, label %common.ret, !dbg !33 br i1 %exec_user_code, label %user_code.entry, label %common.ret, !dbg !33
common.ret: ; preds = %entry, %user_code.entry common.ret: ; preds = %entry, %user_code.entry
ret void, !dbg !34 ret void, !dbg !34
user_code.entry: ; preds = %entry user_code.entry: ; preds = %entry
%1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @9) #3 %1 = call i32 @__kmpc_global_thread_num(%struct.ident_t* nonnull @9) #3
%2 = bitcast [0 x i8*]* %captured_vars_addrs.i2.i to i8* %2 = bitcast [0 x i8*]* %captured_vars_addrs.i2.i to i8*
call void @llvm.lifetime.start.p0i8(i64 0, i8* nonnull %2) #3 call void @llvm.lifetime.start.p0i8(i64 0, i8* nonnull %2) #3
%3 = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @13) #3 %3 = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @13) #3
%4 = getelementptr inbounds [0 x i8*], [0 x i8*]* %captured_vars_addrs.i2.i, i64 0, i64 0, !dbg !35 %4 = getelementptr inbounds [0 x i8*], [0 x i8*]* %captured_vars_addrs.i2.i, i64 0, i64 0, !dbg !35
call void @__kmpc_parallel_51(%struct.ident_t* noundef nonnull @13, i32 %3, i32 noundef 1, i32 noundef -1, i32 noundef -1, i8* noundef bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* noundef bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** noundef nonnull %4, i64 noundef 0) #3, !dbg !35 call void @__kmpc_parallel_51(%struct.ident_t* noundef nonnull @13, i32 %3, i32 noundef 1, i32 noundef -1, i32 noundef -1, i8* noundef bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* noundef bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** noundef nonnull %4, i64 noundef 0) #3, !dbg !35
call void @llvm.lifetime.end.p0i8(i64 0, i8* nonnull %2) #3, !dbg !39 call void @llvm.lifetime.end.p0i8(i64 0, i8* nonnull %2) #3, !dbg !39
call void @llvm.lifetime.start.p0i8(i64 0, i8* nonnull %2) #3 call void @llvm.lifetime.start.p0i8(i64 0, i8* nonnull %2) #3
%5 = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @13) #3 %5 = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @13) #3
call void @__kmpc_parallel_51(%struct.ident_t* noundef nonnull @13, i32 %5, i32 noundef 1, i32 noundef -1, i32 noundef -1, i8* noundef bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* noundef bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** noundef nonnull %4, i64 noundef 0) #3, !dbg !40 call void @__kmpc_parallel_51(%struct.ident_t* noundef nonnull @13, i32 %5, i32 noundef 1, i32 noundef -1, i32 noundef -1, i8* noundef bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* noundef bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** noundef nonnull %4, i64 noundef 0) #3, !dbg !40
call void @llvm.lifetime.end.p0i8(i64 0, i8* nonnull %2) #3, !dbg !42 call void @llvm.lifetime.end.p0i8(i64 0, i8* nonnull %2) #3, !dbg !42
call void @llvm.lifetime.start.p0i8(i64 0, i8* nonnull %2) #3 call void @llvm.lifetime.start.p0i8(i64 0, i8* nonnull %2) #3
%6 = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @13) #3 %6 = call i32 @__kmpc_global_thread_num(%struct.ident_t* noundef nonnull @13) #3
call void @__kmpc_parallel_51(%struct.ident_t* noundef nonnull @13, i32 %6, i32 noundef 1, i32 noundef -1, i32 noundef -1, i8* noundef bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* noundef bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** noundef nonnull %4, i64 noundef 0) #3, !dbg !43 call void @__kmpc_parallel_51(%struct.ident_t* noundef nonnull @13, i32 %6, i32 noundef 1, i32 noundef -1, i32 noundef -1, i8* noundef bitcast (void (i32*, i32*)* @__omp_outlined__2 to i8*), i8* noundef bitcast (void (i16, i32)* @__omp_outlined__2_wrapper to i8*), i8** noundef nonnull %4, i64 noundef 0) #3, !dbg !43
call void @llvm.lifetime.end.p0i8(i64 0, i8* nonnull %2) #3, !dbg !45 call void @llvm.lifetime.end.p0i8(i64 0, i8* nonnull %2) #3, !dbg !45
call void @spmd_amenable() call void @spmd_amenable()
call void @__kmpc_target_deinit(%struct.ident_t* nonnull @11, i8 1, i1 true) #3, !dbg !46 call void @__kmpc_target_deinit(%struct.ident_t* nonnull @11, i8 1) #3, !dbg !46
br label %common.ret br label %common.ret
} }
; Function Attrs: convergent norecurse nounwind ; Function Attrs: convergent norecurse nounwind
define internal void @__omp_outlined__2(i32* noalias nocapture nofree readnone %.global_tid., i32* noalias nocapture nofree readnone %.bound_tid.) #0 !dbg !47 { define internal void @__omp_outlined__2(i32* noalias nocapture nofree readnone %.global_tid., i32* noalias nocapture nofree readnone %.bound_tid.) #0 !dbg !47 {
entry: entry:
call void @unknown() #6, !dbg !48 call void @unknown() #6, !dbg !48
ret void, !dbg !49 ret void, !dbg !49
▲ Show 20 Lines • Show All 92 Lines • Show Last 20 Lines

openmp/libomptarget/DeviceRTL/include/Interface.h

Show First 20 Lines • Show All 211 Lines • ▼ Show 20 Lines
uint32_t __kmpc_get_warp_size(); uint32_t __kmpc_get_warp_size();
/// Kernel /// Kernel
/// ///
///{ ///{
int8_t __kmpc_is_spmd_exec_mode(); int8_t __kmpc_is_spmd_exec_mode();
int32_t __kmpc_target_init(IdentTy *Ident, int8_t Mode, int32_t __kmpc_target_init(IdentTy *Ident, int8_t Mode,
bool UseGenericStateMachine, bool); bool UseGenericStateMachine);
void __kmpc_target_deinit(IdentTy *Ident, int8_t Mode, bool); void __kmpc_target_deinit(IdentTy *Ident, int8_t Mode);
///} ///}
/// Reduction /// Reduction
/// ///
///{ ///{
void __kmpc_nvptx_end_reduce(int32_t TId); void __kmpc_nvptx_end_reduce(int32_t TId);
▲ Show 20 Lines • Show All 126 Lines • Show Last 20 Lines

openmp/libomptarget/DeviceRTL/src/Kernel.cpp

Show First 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
extern "C" { extern "C" {
/// Initialization /// Initialization
/// ///
/// \param Ident Source location identification, can be NULL. /// \param Ident Source location identification, can be NULL.
/// ///
int32_t __kmpc_target_init(IdentTy *Ident, int8_t Mode, int32_t __kmpc_target_init(IdentTy *Ident, int8_t Mode,
bool UseGenericStateMachine, bool) { bool UseGenericStateMachine) {
FunctionTracingRAII(); FunctionTracingRAII();
const bool IsSPMD = Mode & OMP_TGT_EXEC_MODE_SPMD; const bool IsSPMD = Mode & OMP_TGT_EXEC_MODE_SPMD;
if (IsSPMD) { if (IsSPMD) {
inititializeRuntime(/* IsSPMD */ true); inititializeRuntime(/* IsSPMD */ true);
synchronize::threadsAligned(); synchronize::threadsAligned();
} else { } else {
inititializeRuntime(/* IsSPMD */ false); inititializeRuntime(/* IsSPMD */ false);
// No need to wait since only the main threads will execute user // No need to wait since only the main threads will execute user
▲ Show 20 Lines • Show All 42 Lines • ▼ Show 20 Lines
/// De-Initialization /// De-Initialization
/// ///
/// In non-SPMD, this function releases the workers trapped in a state machine /// In non-SPMD, this function releases the workers trapped in a state machine
/// and also any memory dynamically allocated by the runtime. /// and also any memory dynamically allocated by the runtime.
/// ///
/// \param Ident Source location identification, can be NULL. /// \param Ident Source location identification, can be NULL.
/// ///
void __kmpc_target_deinit(IdentTy *Ident, int8_t Mode, bool) { void __kmpc_target_deinit(IdentTy *Ident, int8_t Mode) {
FunctionTracingRAII(); FunctionTracingRAII();
const bool IsSPMD = Mode & OMP_TGT_EXEC_MODE_SPMD; const bool IsSPMD = Mode & OMP_TGT_EXEC_MODE_SPMD;
state::assumeInitialState(IsSPMD); state::assumeInitialState(IsSPMD);
if (IsSPMD) if (IsSPMD)
return; return;
// Signal the workers to exit the state machine and exit the kernel. // Signal the workers to exit the state machine and exit the kernel.
state::ParallelRegionFn = nullptr; state::ParallelRegionFn = nullptr;
Show All 9 Lines