136 lines · cpp
1// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_ size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _ --version 22// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -fopenmp -fopenmp-version=51 -emit-llvm %s -o - | FileCheck --check-prefix=OMP51 %s3// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -fopenmp -fopenmp-version=51 -x c++ -std=c++11 -emit-pch -o %t %s4// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -fopenmp -fopenmp-version=51 -x c++ -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=OMP515 6// expected-no-diagnostics7 8#ifndef HEADER9#define HEADER10 11int thread_limit_target_parallel_for() {12 13// Check that the offloading function is called after setting thread_limit in the task entry function14#pragma omp target parallel for thread_limit(2)15 for(int i=0; i<2; i++) {}16 17 return 0;18}19 20#endif21// OMP51-LABEL: define dso_local noundef i32 @_Z32thread_limit_target_parallel_forv22// OMP51-SAME: () #[[ATTR0:[0-9]+]] {23// OMP51-NEXT: entry:24// OMP51-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 125// OMP51-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2:[0-9]+]])26// OMP51-NEXT: [[TMP1:%.*]] = call ptr @__kmpc_omp_task_alloc(ptr @[[GLOB2]], i32 [[TMP0]], i32 1, i64 40, i64 1, ptr @.omp_task_entry.)27// OMP51-NEXT: [[TMP2:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP1]], i32 0, i32 028// OMP51-NEXT: call void @__kmpc_omp_task_begin_if0(ptr @[[GLOB2]], i32 [[TMP0]], ptr [[TMP1]])29// OMP51-NEXT: [[TMP3:%.*]] = call i32 @.omp_task_entry.(i32 [[TMP0]], ptr [[TMP1]]) #[[ATTR2:[0-9]+]]30// OMP51-NEXT: call void @__kmpc_omp_task_complete_if0(ptr @[[GLOB2]], i32 [[TMP0]], ptr [[TMP1]])31// OMP51-NEXT: ret i32 032//33//34// OMP51-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32thread_limit_target_parallel_forv_l1435// OMP51-SAME: () #[[ATTR1:[0-9]+]] {36// OMP51-NEXT: entry:37// OMP51-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB2]], i32 0, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32thread_limit_target_parallel_forv_l14.omp_outlined)38// OMP51-NEXT: ret void39//40//41// OMP51-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32thread_limit_target_parallel_forv_l14.omp_outlined42// OMP51-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR1]] {43// OMP51-NEXT: entry:44// OMP51-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 845// OMP51-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 846// OMP51-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 447// OMP51-NEXT: [[TMP:%.*]] = alloca i32, align 448// OMP51-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 449// OMP51-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 450// OMP51-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 451// OMP51-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 452// OMP51-NEXT: [[I:%.*]] = alloca i32, align 453// OMP51-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 854// OMP51-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 855// OMP51-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 456// OMP51-NEXT: store i32 1, ptr [[DOTOMP_UB]], align 457// OMP51-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 458// OMP51-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 459// OMP51-NEXT: [[TMP0:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 860// OMP51-NEXT: [[TMP1:%.*]] = load i32, ptr [[TMP0]], align 461// OMP51-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP1]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)62// OMP51-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 463// OMP51-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP2]], 164// OMP51-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]65// OMP51: cond.true:66// OMP51-NEXT: br label [[COND_END:%.*]]67// OMP51: cond.false:68// OMP51-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 469// OMP51-NEXT: br label [[COND_END]]70// OMP51: cond.end:71// OMP51-NEXT: [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP3]], [[COND_FALSE]] ]72// OMP51-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 473// OMP51-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_LB]], align 474// OMP51-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_IV]], align 475// OMP51-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]76// OMP51: omp.inner.for.cond:77// OMP51-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 478// OMP51-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 479// OMP51-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP5]], [[TMP6]]80// OMP51-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]81// OMP51: omp.inner.for.body:82// OMP51-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 483// OMP51-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP7]], 184// OMP51-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]85// OMP51-NEXT: store i32 [[ADD]], ptr [[I]], align 486// OMP51-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]87// OMP51: omp.body.continue:88// OMP51-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]89// OMP51: omp.inner.for.inc:90// OMP51-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 491// OMP51-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP8]], 192// OMP51-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 493// OMP51-NEXT: br label [[OMP_INNER_FOR_COND]]94// OMP51: omp.inner.for.end:95// OMP51-NEXT: br label [[OMP_LOOP_EXIT:%.*]]96// OMP51: omp.loop.exit:97// OMP51-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP1]])98// OMP51-NEXT: ret void99//100//101// OMP51-LABEL: define internal noundef i32 @.omp_task_entry.102// OMP51-SAME: (i32 noundef [[TMP0:%.*]], ptr noalias noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] {103// OMP51-NEXT: entry:104// OMP51-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4105// OMP51-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca ptr, align 8106// OMP51-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca ptr, align 8107// OMP51-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca ptr, align 8108// OMP51-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca ptr, align 8109// OMP51-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca ptr, align 8110// OMP51-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4111// OMP51-NEXT: [[DOTADDR1:%.*]] = alloca ptr, align 8112// OMP51-NEXT: store i32 [[TMP0]], ptr [[DOTADDR]], align 4113// OMP51-NEXT: store ptr [[TMP1]], ptr [[DOTADDR1]], align 8114// OMP51-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTADDR]], align 4115// OMP51-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8116// OMP51-NEXT: [[TMP4:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], ptr [[TMP3]], i32 0, i32 0117// OMP51-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T:%.*]], ptr [[TMP4]], i32 0, i32 2118// OMP51-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT_KMP_TASK_T]], ptr [[TMP4]], i32 0, i32 0119// OMP51-NEXT: [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8120// OMP51-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META5:![0-9]+]])121// OMP51-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META8:![0-9]+]])122// OMP51-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]])123// OMP51-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META12:![0-9]+]])124// OMP51-NEXT: store i32 [[TMP2]], ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META14:![0-9]+]]125// OMP51-NEXT: store ptr [[TMP5]], ptr [[DOTPART_ID__ADDR_I]], align 8, !noalias [[META14]]126// OMP51-NEXT: store ptr null, ptr [[DOTPRIVATES__ADDR_I]], align 8, !noalias [[META14]]127// OMP51-NEXT: store ptr null, ptr [[DOTCOPY_FN__ADDR_I]], align 8, !noalias [[META14]]128// OMP51-NEXT: store ptr [[TMP3]], ptr [[DOTTASK_T__ADDR_I]], align 8, !noalias [[META14]]129// OMP51-NEXT: store ptr [[TMP7]], ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META14]]130// OMP51-NEXT: [[TMP8:%.*]] = load ptr, ptr [[__CONTEXT_ADDR_I]], align 8, !noalias [[META14]]131// OMP51-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias [[META14]]132// OMP51-NEXT: call void @__kmpc_set_thread_limit(ptr @[[GLOB2]], i32 [[TMP9]], i32 2)133// OMP51-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z32thread_limit_target_parallel_forv_l14() #[[ATTR2]]134// OMP51-NEXT: ret i32 0135//136