This is an archive of the discontinued LLVM Phabricator instance.

Differential D120885

[OpenMPOpt] Avoid pointer element type access during region merging
ClosedPublic

Authored by nikic on Mar 3 2022, 2:59 AM.

Details

Reviewers
jdoerfert
ggeorgakoudis
aeubanks
Group Reviewers
Restricted Project
Commits
rG875782bd9ea3: [OpenMPOpt] Avoid pointer element type access during region merging
Summary

Hardcode the function type as ParallelTask, which is the guaranteed pointee type of this runtime function argument (if pointee types exist). The elimination of the callee bitcast is left for InstCombine.

Diff Detail

Event Timeline

nikic created this revision.Mar 3 2022, 2:59 AM
Herald added a project: Restricted Project. · View Herald TranscriptMar 3 2022, 2:59 AM
Herald added subscribers: ormris, guansong, hiraditya, yaxunl. · View Herald Transcript
nikic requested review of this revision.Mar 3 2022, 2:59 AM
Herald added a project: Restricted Project. · View Herald TranscriptMar 3 2022, 2:59 AM
Herald added subscribers: llvm-commits, sstefan1. · View Herald Transcript
Harbormaster completed remote builds in B152325: Diff 412650.Mar 3 2022, 3:55 AM
ormris removed a subscriber: ormris.Mar 7 2022, 9:56 AM
nikic added a reviewer: Restricted Project.Mar 10 2022, 8:31 AM
aeubanks accepted this revision.Mar 14 2022, 9:30 AM
This revision is now accepted and ready to land.Mar 14 2022, 9:30 AM
This revision was landed with ongoing or failed builds.Mar 15 2022, 1:52 AM
Closed by commit rG875782bd9ea3: [OpenMPOpt] Avoid pointer element type access during region merging (authored by nikic). · Explain Why
This revision was automatically updated to reflect the committed changes.
nikic added a commit: rG875782bd9ea3: [OpenMPOpt] Avoid pointer element type access during region merging.

Revision Contents

PathSize
llvm/
lib/
Transforms/
IPO/
6 lines
test/
Transforms/
OpenMP/
96 lines

Diff 415359

llvm/lib/Transforms/IPO/OpenMPOpt.cpp

Show First 20 Lines • Show All 1,111 Lines • ▼ Show 20 Lines auto Merge = [&](const SmallVectorImpl<CallInst *> &MergableCIs,
OMPInfoCache.OMPBuilder.finalize(OriginalFn); OMPInfoCache.OMPBuilder.finalize(OriginalFn);
Function *OutlinedFn = MergableCIs.front()->getCaller(); Function *OutlinedFn = MergableCIs.front()->getCaller();
// Replace the __kmpc_fork_call calls with direct calls to the outlined // Replace the __kmpc_fork_call calls with direct calls to the outlined
// callbacks. // callbacks.
SmallVector<Value *, 8> Args; SmallVector<Value *, 8> Args;
for (auto *CI : MergableCIs) { for (auto *CI : MergableCIs) {
Value *Callee = Value *Callee = CI->getArgOperand(CallbackCalleeOperand);
CI->getArgOperand(CallbackCalleeOperand)->stripPointerCasts(); FunctionType *FT = OMPInfoCache.OMPBuilder.ParallelTask;
FunctionType *FT =
cast<FunctionType>(Callee->getType()->getPointerElementType());
Args.clear(); Args.clear();
Args.push_back(OutlinedFn->getArg(0)); Args.push_back(OutlinedFn->getArg(0));
Args.push_back(OutlinedFn->getArg(1)); Args.push_back(OutlinedFn->getArg(1));
for (unsigned U = CallbackFirstArgOperand, E = CI->arg_size(); U < E; for (unsigned U = CallbackFirstArgOperand, E = CI->arg_size(); U < E;
++U) ++U)
Args.push_back(CI->getArgOperand(U)); Args.push_back(CI->getArgOperand(U));
CallInst *NewCI = CallInst::Create(FT, Callee, Args, "", CI); CallInst *NewCI = CallInst::Create(FT, Callee, Args, "", CI);
▲ Show 20 Lines • Show All 4,075 Lines • Show Last 20 Lines

llvm/test/Transforms/OpenMP/parallel_region_merging.ll

  • This file is larger than 256 KB, so syntax highlighting is disabled by default.
Show First 20 Lines • Show All 4,717 Lines • ▼ Show 20 Lines
; CHECK1-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4 ; CHECK1-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK1: omp.par.region: ; CHECK1: omp.par.region:
; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK1: omp.par.merged: ; CHECK1: omp.par.merged:
; CHECK1-NEXT: call void @.omp_outlined.(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK1-NEXT: call void @.omp_outlined..1(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK1: entry.split: ; CHECK1: entry.split:
; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK1: omp.par.region.split: ; CHECK1: omp.par.region.split:
; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK1: omp.par.pre_finalize: ; CHECK1: omp.par.pre_finalize:
; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK1: omp.par.outlined.exit.exitStub: ; CHECK1: omp.par.outlined.exit.exitStub:
▲ Show 20 Lines • Show All 130 Lines • ▼ Show 20 Lines
; CHECK1-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4 ; CHECK1-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK1: omp.par.region: ; CHECK1: omp.par.region:
; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK1: omp.par.merged: ; CHECK1: omp.par.merged:
; CHECK1-NEXT: call void @.omp_outlined..8(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK1-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 ; CHECK1-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
; CHECK1-NEXT: br i1 [[TMP3]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]] ; CHECK1-NEXT: br i1 [[TMP3]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
; CHECK1: omp_region.end: ; CHECK1: omp_region.end:
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
; CHECK1-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
; CHECK1: omp.par.merged.split.split: ; CHECK1: omp.par.merged.split.split:
; CHECK1-NEXT: call void @.omp_outlined..9(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..9 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK1: entry.split: ; CHECK1: entry.split:
; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK1: omp.par.region.split: ; CHECK1: omp.par.region.split:
; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK1: omp.par.pre_finalize: ; CHECK1: omp.par.pre_finalize:
; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK1: omp_region.body: ; CHECK1: omp_region.body:
▲ Show 20 Lines • Show All 68 Lines • ▼ Show 20 Lines
; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: [[TMP2:%.*]] = load float, float* [[LOADGEP_F_RELOADED]], align 4 ; CHECK1-NEXT: [[TMP2:%.*]] = load float, float* [[LOADGEP_F_RELOADED]], align 4
; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK1: omp.par.region: ; CHECK1: omp.par.region:
; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK1: omp.par.merged: ; CHECK1: omp.par.merged:
; CHECK1-NEXT: call void @.omp_outlined..10(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_F_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, float*)* @.omp_outlined..10 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_F_ADDR]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK1-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK1-NEXT: [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0 ; CHECK1-NEXT: [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0
; CHECK1-NEXT: br i1 [[TMP4]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]] ; CHECK1-NEXT: br i1 [[TMP4]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
; CHECK1: omp_region.end: ; CHECK1: omp_region.end:
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
; CHECK1-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
; CHECK1: omp.par.merged.split.split: ; CHECK1: omp.par.merged.split.split:
; CHECK1-NEXT: call void @.omp_outlined..11(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_F_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, float*)* @.omp_outlined..11 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_F_ADDR]])
; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK1: entry.split: ; CHECK1: entry.split:
; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK1: omp.par.region.split: ; CHECK1: omp.par.region.split:
; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK1: omp.par.pre_finalize: ; CHECK1: omp.par.pre_finalize:
; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK1: omp_region.body: ; CHECK1: omp_region.body:
▲ Show 20 Lines • Show All 65 Lines • ▼ Show 20 Lines
; CHECK1-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4 ; CHECK1-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK1: omp.par.region: ; CHECK1: omp.par.region:
; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK1: omp.par.merged: ; CHECK1: omp.par.merged:
; CHECK1-NEXT: call void @.omp_outlined..12(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..12 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK1-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 ; CHECK1-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
; CHECK1-NEXT: br i1 [[TMP3]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]] ; CHECK1-NEXT: br i1 [[TMP3]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
; CHECK1: omp_region.end: ; CHECK1: omp_region.end:
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
; CHECK1-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
; CHECK1: omp.par.merged.split.split: ; CHECK1: omp.par.merged.split.split:
; CHECK1-NEXT: [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD:%.*]] = load i64, i64* [[LOADGEP_A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]], align 8 ; CHECK1-NEXT: [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD:%.*]] = load i64, i64* [[LOADGEP_A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]], align 8
; CHECK1-NEXT: call void @.omp_outlined..13(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i64 [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i64)* @.omp_outlined..13 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i64 [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD]])
; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK1: entry.split: ; CHECK1: entry.split:
; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK1: omp.par.region.split: ; CHECK1: omp.par.region.split:
; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK1: omp.par.pre_finalize: ; CHECK1: omp.par.pre_finalize:
; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK1: omp_region.body: ; CHECK1: omp_region.body:
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
; CHECK1-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4 ; CHECK1-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK1: omp.par.region: ; CHECK1: omp.par.region:
; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK1: omp.par.merged: ; CHECK1: omp.par.merged:
; CHECK1-NEXT: call void @.omp_outlined..14(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..14 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK1-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK1-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 ; CHECK1-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
; CHECK1-NEXT: br i1 [[TMP3]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]] ; CHECK1-NEXT: br i1 [[TMP3]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
; CHECK1: omp_region.end: ; CHECK1: omp_region.end:
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
; CHECK1-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
; CHECK1: omp.par.merged.split.split: ; CHECK1: omp.par.merged.split.split:
; CHECK1-NEXT: call void @.omp_outlined..15(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..15 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK1: entry.split: ; CHECK1: entry.split:
; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK1: omp.par.region.split: ; CHECK1: omp.par.region.split:
; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK1: omp.par.pre_finalize: ; CHECK1: omp.par.pre_finalize:
; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK1: omp_region.body: ; CHECK1: omp_region.body:
▲ Show 20 Lines • Show All 76 Lines • ▼ Show 20 Lines
; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[LOADGEP_A_RELOADED]], align 4 ; CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[LOADGEP_A_RELOADED]], align 4
; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK1: omp.par.region: ; CHECK1: omp.par.region:
; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK1: omp.par.merged: ; CHECK1: omp.par.merged:
; CHECK1-NEXT: call void @.omp_outlined..16(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..16 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK1-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK1-NEXT: [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0 ; CHECK1-NEXT: [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0
; CHECK1-NEXT: br i1 [[TMP4]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]] ; CHECK1-NEXT: br i1 [[TMP4]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
; CHECK1: omp_region.end: ; CHECK1: omp_region.end:
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
; CHECK1-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
; CHECK1: omp.par.merged.split.split: ; CHECK1: omp.par.merged.split.split:
; CHECK1-NEXT: call void @.omp_outlined..17(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_B]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*, i32*)* @.omp_outlined..17 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_B]])
; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK1: entry.split: ; CHECK1: entry.split:
; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK1: omp.par.region.split: ; CHECK1: omp.par.region.split:
; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK1: omp.par.pre_finalize: ; CHECK1: omp.par.pre_finalize:
; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK1: omp_region.body: ; CHECK1: omp_region.body:
▲ Show 20 Lines • Show All 65 Lines • ▼ Show 20 Lines
; CHECK1-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4 ; CHECK1-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK1: omp.par.region: ; CHECK1: omp.par.region:
; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK1: omp.par.merged: ; CHECK1: omp.par.merged:
; CHECK1-NEXT: call void @.omp_outlined..18(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_CANCEL1_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..18 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_CANCEL1_ADDR]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK1-NEXT: call void @.omp_outlined..19(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_CANCEL2_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..19 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_CANCEL2_ADDR]])
; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK1: entry.split: ; CHECK1: entry.split:
; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK1: omp.par.region.split: ; CHECK1: omp.par.region.split:
; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK1: omp.par.pre_finalize: ; CHECK1: omp.par.pre_finalize:
; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK1: omp.par.outlined.exit.exitStub: ; CHECK1: omp.par.outlined.exit.exitStub:
▲ Show 20 Lines • Show All 70 Lines • ▼ Show 20 Lines
; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[LOADGEP_CANCEL1_RELOADED]], align 4 ; CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[LOADGEP_CANCEL1_RELOADED]], align 4
; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK1: omp.par.region: ; CHECK1: omp.par.region:
; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK1: omp.par.merged: ; CHECK1: omp.par.merged:
; CHECK1-NEXT: call void @.omp_outlined..20(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_CANCEL1_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..20 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_CANCEL1_ADDR]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK1-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK1-NEXT: [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0 ; CHECK1-NEXT: [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0
; CHECK1-NEXT: br i1 [[TMP4]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]] ; CHECK1-NEXT: br i1 [[TMP4]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
; CHECK1: omp_region.end: ; CHECK1: omp_region.end:
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
; CHECK1-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
; CHECK1: omp.par.merged.split.split: ; CHECK1: omp.par.merged.split.split:
; CHECK1-NEXT: call void @.omp_outlined..21(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_CANCEL2_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..21 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_CANCEL2_ADDR]])
; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK1: entry.split: ; CHECK1: entry.split:
; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK1: omp.par.region.split: ; CHECK1: omp.par.region.split:
; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK1: omp.par.pre_finalize: ; CHECK1: omp.par.pre_finalize:
; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK1: omp_region.body: ; CHECK1: omp_region.body:
▲ Show 20 Lines • Show All 69 Lines • ▼ Show 20 Lines
; CHECK1-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4 ; CHECK1-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK1: omp.par.region: ; CHECK1: omp.par.region:
; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK1: omp.par.merged: ; CHECK1: omp.par.merged:
; CHECK1-NEXT: call void @.omp_outlined..22(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..22 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK1-NEXT: call void @.omp_outlined..23(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..23 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
; CHECK1-NEXT: call void @.omp_outlined..24(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..24 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK1: entry.split: ; CHECK1: entry.split:
; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK1: omp.par.region.split: ; CHECK1: omp.par.region.split:
; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK1: omp.par.pre_finalize: ; CHECK1: omp.par.pre_finalize:
; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK1: omp.par.outlined.exit.exitStub: ; CHECK1: omp.par.outlined.exit.exitStub:
▲ Show 20 Lines • Show All 72 Lines • ▼ Show 20 Lines
; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[LOADGEP_A_RELOADED]], align 4 ; CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[LOADGEP_A_RELOADED]], align 4
; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK1: omp.par.region: ; CHECK1: omp.par.region:
; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK1: omp.par.merged: ; CHECK1: omp.par.merged:
; CHECK1-NEXT: call void @.omp_outlined..25(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..25 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK1-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK1-NEXT: [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0 ; CHECK1-NEXT: [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0
; CHECK1-NEXT: br i1 [[TMP4]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]] ; CHECK1-NEXT: br i1 [[TMP4]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
; CHECK1: omp_region.end: ; CHECK1: omp_region.end:
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
; CHECK1-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
; CHECK1: omp.par.merged.split.split: ; CHECK1: omp.par.merged.split.split:
; CHECK1-NEXT: call void @.omp_outlined..26(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..26 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM4:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM4:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM4]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM4]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM3:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM3:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: [[TMP5:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM3]]) ; CHECK1-NEXT: [[TMP5:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM3]])
; CHECK1-NEXT: [[TMP6:%.*]] = icmp ne i32 [[TMP5]], 0 ; CHECK1-NEXT: [[TMP6:%.*]] = icmp ne i32 [[TMP5]], 0
; CHECK1-NEXT: br i1 [[TMP6]], label [[OMP_REGION_BODY5:%.*]], label [[OMP_REGION_END4:%.*]] ; CHECK1-NEXT: br i1 [[TMP6]], label [[OMP_REGION_BODY5:%.*]], label [[OMP_REGION_END4:%.*]]
; CHECK1: omp_region.end4: ; CHECK1: omp_region.end4:
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM6:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM6:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM6]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM6]])
; CHECK1-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT_SPLIT_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT_SPLIT_SPLIT:%.*]]
; CHECK1: omp.par.merged.split.split.split.split: ; CHECK1: omp.par.merged.split.split.split.split:
; CHECK1-NEXT: call void @.omp_outlined..27(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..27 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK1: entry.split: ; CHECK1: entry.split:
; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK1: omp.par.region.split: ; CHECK1: omp.par.region.split:
; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK1: omp.par.pre_finalize: ; CHECK1: omp.par.pre_finalize:
; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK1: omp_region.body5: ; CHECK1: omp_region.body5:
▲ Show 20 Lines • Show All 189 Lines • ▼ Show 20 Lines
; CHECK1-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4 ; CHECK1-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK1-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK1: omp.par.region: ; CHECK1: omp.par.region:
; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK1: omp.par.merged: ; CHECK1: omp.par.merged:
; CHECK1-NEXT: call void @.omp_outlined..37(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..37 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK1-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK1-NEXT: call void @.omp_outlined..38(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK1-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..38 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK1-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK1: entry.split: ; CHECK1: entry.split:
; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK1: omp.par.region.split: ; CHECK1: omp.par.region.split:
; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK1: omp.par.pre_finalize: ; CHECK1: omp.par.pre_finalize:
; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK1-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK1: omp.par.outlined.exit.exitStub: ; CHECK1: omp.par.outlined.exit.exitStub:
▲ Show 20 Lines • Show All 53 Lines • ▼ Show 20 Lines
; CHECK2-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4 ; CHECK2-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK2: omp.par.region: ; CHECK2: omp.par.region:
; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK2: omp.par.merged: ; CHECK2: omp.par.merged:
; CHECK2-NEXT: call void @.omp_outlined.(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK2-NEXT: call void @.omp_outlined..1(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK2: entry.split: ; CHECK2: entry.split:
; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK2: omp.par.region.split: ; CHECK2: omp.par.region.split:
; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK2: omp.par.pre_finalize: ; CHECK2: omp.par.pre_finalize:
; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK2: omp.par.outlined.exit.exitStub: ; CHECK2: omp.par.outlined.exit.exitStub:
▲ Show 20 Lines • Show All 130 Lines • ▼ Show 20 Lines
; CHECK2-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4 ; CHECK2-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK2: omp.par.region: ; CHECK2: omp.par.region:
; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK2: omp.par.merged: ; CHECK2: omp.par.merged:
; CHECK2-NEXT: call void @.omp_outlined..8(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..8 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK2-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 ; CHECK2-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
; CHECK2-NEXT: br i1 [[TMP3]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]] ; CHECK2-NEXT: br i1 [[TMP3]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
; CHECK2: omp_region.end: ; CHECK2: omp_region.end:
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
; CHECK2-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
; CHECK2: omp.par.merged.split.split: ; CHECK2: omp.par.merged.split.split:
; CHECK2-NEXT: call void @.omp_outlined..9(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..9 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK2: entry.split: ; CHECK2: entry.split:
; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK2: omp.par.region.split: ; CHECK2: omp.par.region.split:
; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK2: omp.par.pre_finalize: ; CHECK2: omp.par.pre_finalize:
; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK2: omp_region.body: ; CHECK2: omp_region.body:
▲ Show 20 Lines • Show All 68 Lines • ▼ Show 20 Lines
; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: [[TMP2:%.*]] = load float, float* [[LOADGEP_F_RELOADED]], align 4 ; CHECK2-NEXT: [[TMP2:%.*]] = load float, float* [[LOADGEP_F_RELOADED]], align 4
; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK2: omp.par.region: ; CHECK2: omp.par.region:
; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK2: omp.par.merged: ; CHECK2: omp.par.merged:
; CHECK2-NEXT: call void @.omp_outlined..10(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_F_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, float*)* @.omp_outlined..10 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_F_ADDR]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK2-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK2-NEXT: [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0 ; CHECK2-NEXT: [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0
; CHECK2-NEXT: br i1 [[TMP4]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]] ; CHECK2-NEXT: br i1 [[TMP4]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
; CHECK2: omp_region.end: ; CHECK2: omp_region.end:
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
; CHECK2-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
; CHECK2: omp.par.merged.split.split: ; CHECK2: omp.par.merged.split.split:
; CHECK2-NEXT: call void @.omp_outlined..11(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_F_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, float*)* @.omp_outlined..11 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], float* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_F_ADDR]])
; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK2: entry.split: ; CHECK2: entry.split:
; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK2: omp.par.region.split: ; CHECK2: omp.par.region.split:
; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK2: omp.par.pre_finalize: ; CHECK2: omp.par.pre_finalize:
; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK2: omp_region.body: ; CHECK2: omp_region.body:
▲ Show 20 Lines • Show All 65 Lines • ▼ Show 20 Lines
; CHECK2-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4 ; CHECK2-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK2: omp.par.region: ; CHECK2: omp.par.region:
; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK2: omp.par.merged: ; CHECK2: omp.par.merged:
; CHECK2-NEXT: call void @.omp_outlined..12(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..12 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK2-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 ; CHECK2-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
; CHECK2-NEXT: br i1 [[TMP3]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]] ; CHECK2-NEXT: br i1 [[TMP3]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
; CHECK2: omp_region.end: ; CHECK2: omp_region.end:
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
; CHECK2-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
; CHECK2: omp.par.merged.split.split: ; CHECK2: omp.par.merged.split.split:
; CHECK2-NEXT: [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD:%.*]] = load i64, i64* [[LOADGEP_A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]], align 8 ; CHECK2-NEXT: [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD:%.*]] = load i64, i64* [[LOADGEP_A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_ALLOC]], align 8
; CHECK2-NEXT: call void @.omp_outlined..13(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i64 [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i64)* @.omp_outlined..13 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i64 [[A_CASTED_SROA_0_0_INSERT_EXT_SEQ_OUTPUT_LOAD]])
; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK2: entry.split: ; CHECK2: entry.split:
; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK2: omp.par.region.split: ; CHECK2: omp.par.region.split:
; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK2: omp.par.pre_finalize: ; CHECK2: omp.par.pre_finalize:
; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK2: omp_region.body: ; CHECK2: omp_region.body:
▲ Show 20 Lines • Show All 60 Lines • ▼ Show 20 Lines
; CHECK2-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4 ; CHECK2-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK2: omp.par.region: ; CHECK2: omp.par.region:
; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK2: omp.par.merged: ; CHECK2: omp.par.merged:
; CHECK2-NEXT: call void @.omp_outlined..14(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..14 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK2-NEXT: [[TMP2:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK2-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0 ; CHECK2-NEXT: [[TMP3:%.*]] = icmp ne i32 [[TMP2]], 0
; CHECK2-NEXT: br i1 [[TMP3]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]] ; CHECK2-NEXT: br i1 [[TMP3]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
; CHECK2: omp_region.end: ; CHECK2: omp_region.end:
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
; CHECK2-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
; CHECK2: omp.par.merged.split.split: ; CHECK2: omp.par.merged.split.split:
; CHECK2-NEXT: call void @.omp_outlined..15(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..15 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK2: entry.split: ; CHECK2: entry.split:
; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK2: omp.par.region.split: ; CHECK2: omp.par.region.split:
; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK2: omp.par.pre_finalize: ; CHECK2: omp.par.pre_finalize:
; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK2: omp_region.body: ; CHECK2: omp_region.body:
▲ Show 20 Lines • Show All 76 Lines • ▼ Show 20 Lines
; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[LOADGEP_A_RELOADED]], align 4 ; CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[LOADGEP_A_RELOADED]], align 4
; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK2: omp.par.region: ; CHECK2: omp.par.region:
; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK2: omp.par.merged: ; CHECK2: omp.par.merged:
; CHECK2-NEXT: call void @.omp_outlined..16(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..16 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK2-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK2-NEXT: [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0 ; CHECK2-NEXT: [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0
; CHECK2-NEXT: br i1 [[TMP4]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]] ; CHECK2-NEXT: br i1 [[TMP4]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
; CHECK2: omp_region.end: ; CHECK2: omp_region.end:
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
; CHECK2-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
; CHECK2: omp.par.merged.split.split: ; CHECK2: omp.par.merged.split.split:
; CHECK2-NEXT: call void @.omp_outlined..17(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_B]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*, i32*)* @.omp_outlined..17 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_B]])
; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK2: entry.split: ; CHECK2: entry.split:
; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK2: omp.par.region.split: ; CHECK2: omp.par.region.split:
; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK2: omp.par.pre_finalize: ; CHECK2: omp.par.pre_finalize:
; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK2: omp_region.body: ; CHECK2: omp_region.body:
▲ Show 20 Lines • Show All 65 Lines • ▼ Show 20 Lines
; CHECK2-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4 ; CHECK2-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK2: omp.par.region: ; CHECK2: omp.par.region:
; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK2: omp.par.merged: ; CHECK2: omp.par.merged:
; CHECK2-NEXT: call void @.omp_outlined..18(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_CANCEL1_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..18 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_CANCEL1_ADDR]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK2-NEXT: call void @.omp_outlined..19(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_CANCEL2_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..19 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_CANCEL2_ADDR]])
; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK2: entry.split: ; CHECK2: entry.split:
; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK2: omp.par.region.split: ; CHECK2: omp.par.region.split:
; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK2: omp.par.pre_finalize: ; CHECK2: omp.par.pre_finalize:
; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK2: omp.par.outlined.exit.exitStub: ; CHECK2: omp.par.outlined.exit.exitStub:
▲ Show 20 Lines • Show All 70 Lines • ▼ Show 20 Lines
; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[LOADGEP_CANCEL1_RELOADED]], align 4 ; CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[LOADGEP_CANCEL1_RELOADED]], align 4
; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK2: omp.par.region: ; CHECK2: omp.par.region:
; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK2: omp.par.merged: ; CHECK2: omp.par.merged:
; CHECK2-NEXT: call void @.omp_outlined..20(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_CANCEL1_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..20 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_CANCEL1_ADDR]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK2-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK2-NEXT: [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0 ; CHECK2-NEXT: [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0
; CHECK2-NEXT: br i1 [[TMP4]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]] ; CHECK2-NEXT: br i1 [[TMP4]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
; CHECK2: omp_region.end: ; CHECK2: omp_region.end:
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
; CHECK2-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
; CHECK2: omp.par.merged.split.split: ; CHECK2: omp.par.merged.split.split:
; CHECK2-NEXT: call void @.omp_outlined..21(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_CANCEL2_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..21 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_CANCEL2_ADDR]])
; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK2: entry.split: ; CHECK2: entry.split:
; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK2: omp.par.region.split: ; CHECK2: omp.par.region.split:
; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK2: omp.par.pre_finalize: ; CHECK2: omp.par.pre_finalize:
; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK2: omp_region.body: ; CHECK2: omp_region.body:
▲ Show 20 Lines • Show All 69 Lines • ▼ Show 20 Lines
; CHECK2-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4 ; CHECK2-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK2: omp.par.region: ; CHECK2: omp.par.region:
; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK2: omp.par.merged: ; CHECK2: omp.par.merged:
; CHECK2-NEXT: call void @.omp_outlined..22(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..22 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK2-NEXT: call void @.omp_outlined..23(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..23 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
; CHECK2-NEXT: call void @.omp_outlined..24(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..24 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK2: entry.split: ; CHECK2: entry.split:
; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK2: omp.par.region.split: ; CHECK2: omp.par.region.split:
; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK2: omp.par.pre_finalize: ; CHECK2: omp.par.pre_finalize:
; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK2: omp.par.outlined.exit.exitStub: ; CHECK2: omp.par.outlined.exit.exitStub:
▲ Show 20 Lines • Show All 72 Lines • ▼ Show 20 Lines
; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[LOADGEP_A_RELOADED]], align 4 ; CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[LOADGEP_A_RELOADED]], align 4
; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK2: omp.par.region: ; CHECK2: omp.par.region:
; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK2: omp.par.merged: ; CHECK2: omp.par.merged:
; CHECK2-NEXT: call void @.omp_outlined..25(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..25 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM2]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK2-NEXT: [[TMP3:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK2-NEXT: [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0 ; CHECK2-NEXT: [[TMP4:%.*]] = icmp ne i32 [[TMP3]], 0
; CHECK2-NEXT: br i1 [[TMP4]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]] ; CHECK2-NEXT: br i1 [[TMP4]], label [[OMP_REGION_BODY:%.*]], label [[OMP_REGION_END:%.*]]
; CHECK2: omp_region.end: ; CHECK2: omp_region.end:
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM1:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM1]])
; CHECK2-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT:%.*]]
; CHECK2: omp.par.merged.split.split: ; CHECK2: omp.par.merged.split.split:
; CHECK2-NEXT: call void @.omp_outlined..26(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..26 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM4:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM4:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM4]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM4]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM3:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM3:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: [[TMP5:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM3]]) ; CHECK2-NEXT: [[TMP5:%.*]] = call i32 @__kmpc_master(%struct.ident_t* @[[GLOB2]], i32 [[OMP_GLOBAL_THREAD_NUM3]])
; CHECK2-NEXT: [[TMP6:%.*]] = icmp ne i32 [[TMP5]], 0 ; CHECK2-NEXT: [[TMP6:%.*]] = icmp ne i32 [[TMP5]], 0
; CHECK2-NEXT: br i1 [[TMP6]], label [[OMP_REGION_BODY5:%.*]], label [[OMP_REGION_END4:%.*]] ; CHECK2-NEXT: br i1 [[TMP6]], label [[OMP_REGION_BODY5:%.*]], label [[OMP_REGION_END4:%.*]]
; CHECK2: omp_region.end4: ; CHECK2: omp_region.end4:
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM6:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM6:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM6]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM6]])
; CHECK2-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT_SPLIT_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED_SPLIT_SPLIT_SPLIT_SPLIT:%.*]]
; CHECK2: omp.par.merged.split.split.split.split: ; CHECK2: omp.par.merged.split.split.split.split:
; CHECK2-NEXT: call void @.omp_outlined..27(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..27 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK2: entry.split: ; CHECK2: entry.split:
; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK2: omp.par.region.split: ; CHECK2: omp.par.region.split:
; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK2: omp.par.pre_finalize: ; CHECK2: omp.par.pre_finalize:
; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK2: omp_region.body5: ; CHECK2: omp_region.body5:
▲ Show 20 Lines • Show All 189 Lines • ▼ Show 20 Lines
; CHECK2-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4 ; CHECK2-NEXT: [[TID_ADDR_LOCAL:%.*]] = alloca i32, align 4
; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4 ; CHECK2-NEXT: [[TMP1:%.*]] = load i32, i32* [[TID_ADDR]], align 4
; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: store i32 [[TMP1]], i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4 ; CHECK2-NEXT: [[TID:%.*]] = load i32, i32* [[TID_ADDR_LOCAL]], align 4
; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION:%.*]]
; CHECK2: omp.par.region: ; CHECK2: omp.par.region:
; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_MERGED:%.*]]
; CHECK2: omp.par.merged: ; CHECK2: omp.par.merged:
; CHECK2-NEXT: call void @.omp_outlined..37(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..37 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) ; CHECK2-NEXT: [[OMP_GLOBAL_THREAD_NUM:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]])
; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM]]) ; CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB3]], i32 [[OMP_GLOBAL_THREAD_NUM]])
; CHECK2-NEXT: call void @.omp_outlined..38(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]]) ; CHECK2-NEXT: call void (i32*, i32*, ...) bitcast (void (i32*, i32*, i32*)* @.omp_outlined..38 to void (i32*, i32*, ...)*)(i32* [[TID_ADDR]], i32* [[ZERO_ADDR]], i32* nocapture nofree noundef nonnull readonly align 4 dereferenceable(4) [[LOADGEP_A_ADDR]])
; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]] ; CHECK2-NEXT: br label [[ENTRY_SPLIT:%.*]]
; CHECK2: entry.split: ; CHECK2: entry.split:
; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_REGION_SPLIT:%.*]]
; CHECK2: omp.par.region.split: ; CHECK2: omp.par.region.split:
; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_PRE_FINALIZE:%.*]]
; CHECK2: omp.par.pre_finalize: ; CHECK2: omp.par.pre_finalize:
; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]] ; CHECK2-NEXT: br label [[OMP_PAR_OUTLINED_EXIT_EXITSTUB:%.*]]
; CHECK2: omp.par.outlined.exit.exitStub: ; CHECK2: omp.par.outlined.exit.exitStub:
Show All 26 Lines