572 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// Test host codegen.3// RUN: %clang_cc1 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s4// expected-no-diagnostics5 6int kernel_within_loop(int *a, int *b, int N, int num_iters) {7 int i;8 for (i = 0; i < num_iters; ++i) {9#pragma omp target parallel for map(a[0:N]) map(b[0:N])10 for (int j = 0; j< N; j++)11 a[j] = b[j];12 13#pragma omp target teams distribute parallel for map(a[0:N]) map(b[0:N])14 for (int j = 0; j< N; j+=3)15 a[j] = b[j] * 2;16 }17 return a[N-1];18}19// CHECK-LABEL: define dso_local noundef signext i32 @_Z18kernel_within_loopPiS_ii20// CHECK-SAME: (ptr noundef [[A:%.*]], ptr noundef [[B:%.*]], i32 noundef signext [[N:%.*]], i32 noundef signext [[NUM_ITERS:%.*]]) #[[ATTR0:[0-9]+]] {21// CHECK-NEXT: entry:22// CHECK-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 823// CHECK-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 824// CHECK-NEXT: [[N_ADDR:%.*]] = alloca i32, align 425// CHECK-NEXT: [[NUM_ITERS_ADDR:%.*]] = alloca i32, align 426// CHECK-NEXT: [[I:%.*]] = alloca i32, align 427// CHECK-NEXT: [[N_CASTED:%.*]] = alloca i64, align 828// CHECK-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 829// CHECK-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 830// CHECK-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 831// CHECK-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 832// CHECK-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 833// CHECK-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 834// CHECK-NEXT: [[DOTOFFLOAD_BASEPTRS8:%.*]] = alloca [3 x ptr], align 835// CHECK-NEXT: [[DOTOFFLOAD_PTRS9:%.*]] = alloca [3 x ptr], align 836// CHECK-NEXT: [[DOTOFFLOAD_MAPPERS10:%.*]] = alloca [3 x ptr], align 837// CHECK-NEXT: [[DOTOFFLOAD_SIZES11:%.*]] = alloca [3 x i64], align 838// CHECK-NEXT: [[TMP:%.*]] = alloca i32, align 439// CHECK-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 440// CHECK-NEXT: [[DOTCAPTURE_EXPR_12:%.*]] = alloca i32, align 441// CHECK-NEXT: [[KERNEL_ARGS14:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 842// CHECK-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 843// CHECK-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 844// CHECK-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 445// CHECK-NEXT: store i32 [[NUM_ITERS]], ptr [[NUM_ITERS_ADDR]], align 446// CHECK-NEXT: store i32 0, ptr [[I]], align 447// CHECK-NEXT: br label [[FOR_COND:%.*]]48// CHECK: for.cond:49// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[I]], align 450// CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[NUM_ITERS_ADDR]], align 451// CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 [[TMP0]], [[TMP1]]52// CHECK-NEXT: br i1 [[CMP]], label [[FOR_BODY:%.*]], label [[FOR_END:%.*]]53// CHECK: for.body:54// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 455// CHECK-NEXT: store i32 [[TMP2]], ptr [[N_CASTED]], align 456// CHECK-NEXT: [[TMP3:%.*]] = load i64, ptr [[N_CASTED]], align 857// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[A_ADDR]], align 858// CHECK-NEXT: [[TMP5:%.*]] = load ptr, ptr [[B_ADDR]], align 859// CHECK-NEXT: [[TMP6:%.*]] = load ptr, ptr [[A_ADDR]], align 860// CHECK-NEXT: [[TMP7:%.*]] = load ptr, ptr [[A_ADDR]], align 861// CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP7]], i64 062// CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[N_ADDR]], align 463// CHECK-NEXT: [[CONV:%.*]] = sext i32 [[TMP8]] to i6464// CHECK-NEXT: [[TMP9:%.*]] = mul nuw i64 [[CONV]], 465// CHECK-NEXT: [[TMP10:%.*]] = load ptr, ptr [[B_ADDR]], align 866// CHECK-NEXT: [[TMP11:%.*]] = load ptr, ptr [[B_ADDR]], align 867// CHECK-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP11]], i64 068// CHECK-NEXT: [[TMP12:%.*]] = load i32, ptr [[N_ADDR]], align 469// CHECK-NEXT: [[CONV2:%.*]] = sext i32 [[TMP12]] to i6470// CHECK-NEXT: [[TMP13:%.*]] = mul nuw i64 [[CONV2]], 471// CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 24, i1 false)72// CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 073// CHECK-NEXT: store i64 [[TMP3]], ptr [[TMP14]], align 874// CHECK-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 075// CHECK-NEXT: store i64 [[TMP3]], ptr [[TMP15]], align 876// CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 077// CHECK-NEXT: store ptr null, ptr [[TMP16]], align 878// CHECK-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 179// CHECK-NEXT: store ptr [[TMP6]], ptr [[TMP17]], align 880// CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 181// CHECK-NEXT: store ptr [[ARRAYIDX]], ptr [[TMP18]], align 882// CHECK-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 183// CHECK-NEXT: store i64 [[TMP9]], ptr [[TMP19]], align 884// CHECK-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 185// CHECK-NEXT: store ptr null, ptr [[TMP20]], align 886// CHECK-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 287// CHECK-NEXT: store ptr [[TMP10]], ptr [[TMP21]], align 888// CHECK-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 289// CHECK-NEXT: store ptr [[ARRAYIDX1]], ptr [[TMP22]], align 890// CHECK-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 291// CHECK-NEXT: store i64 [[TMP13]], ptr [[TMP23]], align 892// CHECK-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 293// CHECK-NEXT: store ptr null, ptr [[TMP24]], align 894// CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 095// CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 096// CHECK-NEXT: [[TMP27:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 097// CHECK-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 098// CHECK-NEXT: store i32 3, ptr [[TMP28]], align 499// CHECK-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1100// CHECK-NEXT: store i32 3, ptr [[TMP29]], align 4101// CHECK-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2102// CHECK-NEXT: store ptr [[TMP25]], ptr [[TMP30]], align 8103// CHECK-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3104// CHECK-NEXT: store ptr [[TMP26]], ptr [[TMP31]], align 8105// CHECK-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4106// CHECK-NEXT: store ptr [[TMP27]], ptr [[TMP32]], align 8107// CHECK-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5108// CHECK-NEXT: store ptr @.offload_maptypes, ptr [[TMP33]], align 8109// CHECK-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6110// CHECK-NEXT: store ptr null, ptr [[TMP34]], align 8111// CHECK-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7112// CHECK-NEXT: store ptr null, ptr [[TMP35]], align 8113// CHECK-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8114// CHECK-NEXT: store i64 0, ptr [[TMP36]], align 8115// CHECK-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9116// CHECK-NEXT: store i64 0, ptr [[TMP37]], align 8117// CHECK-NEXT: [[TMP38:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10118// CHECK-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP38]], align 4119// CHECK-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11120// CHECK-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP39]], align 4121// CHECK-NEXT: [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12122// CHECK-NEXT: store i32 0, ptr [[TMP40]], align 4123// CHECK-NEXT: [[TMP41:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 1, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18kernel_within_loopPiS_ii_l9.region_id, ptr [[KERNEL_ARGS]])124// CHECK-NEXT: [[TMP42:%.*]] = icmp ne i32 [[TMP41]], 0125// CHECK-NEXT: br i1 [[TMP42]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]]126// CHECK: omp_offload.failed:127// CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18kernel_within_loopPiS_ii_l9(i64 [[TMP3]], ptr [[TMP4]], ptr [[TMP5]]) #[[ATTR2:[0-9]+]]128// CHECK-NEXT: br label [[OMP_OFFLOAD_CONT]]129// CHECK: omp_offload.cont:130// CHECK-NEXT: [[TMP43:%.*]] = load i32, ptr [[N_ADDR]], align 4131// CHECK-NEXT: store i32 [[TMP43]], ptr [[N_CASTED3]], align 4132// CHECK-NEXT: [[TMP44:%.*]] = load i64, ptr [[N_CASTED3]], align 8133// CHECK-NEXT: [[TMP45:%.*]] = load ptr, ptr [[A_ADDR]], align 8134// CHECK-NEXT: [[TMP46:%.*]] = load ptr, ptr [[B_ADDR]], align 8135// CHECK-NEXT: [[TMP47:%.*]] = load ptr, ptr [[A_ADDR]], align 8136// CHECK-NEXT: [[TMP48:%.*]] = load ptr, ptr [[A_ADDR]], align 8137// CHECK-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP48]], i64 0138// CHECK-NEXT: [[TMP49:%.*]] = load i32, ptr [[N_ADDR]], align 4139// CHECK-NEXT: [[CONV5:%.*]] = sext i32 [[TMP49]] to i64140// CHECK-NEXT: [[TMP50:%.*]] = mul nuw i64 [[CONV5]], 4141// CHECK-NEXT: [[TMP51:%.*]] = load ptr, ptr [[B_ADDR]], align 8142// CHECK-NEXT: [[TMP52:%.*]] = load ptr, ptr [[B_ADDR]], align 8143// CHECK-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds nuw i32, ptr [[TMP52]], i64 0144// CHECK-NEXT: [[TMP53:%.*]] = load i32, ptr [[N_ADDR]], align 4145// CHECK-NEXT: [[CONV7:%.*]] = sext i32 [[TMP53]] to i64146// CHECK-NEXT: [[TMP54:%.*]] = mul nuw i64 [[CONV7]], 4147// CHECK-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES11]], ptr align 8 @.offload_sizes.1, i64 24, i1 false)148// CHECK-NEXT: [[TMP55:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0149// CHECK-NEXT: store i64 [[TMP44]], ptr [[TMP55]], align 8150// CHECK-NEXT: [[TMP56:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0151// CHECK-NEXT: store i64 [[TMP44]], ptr [[TMP56]], align 8152// CHECK-NEXT: [[TMP57:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i64 0, i64 0153// CHECK-NEXT: store ptr null, ptr [[TMP57]], align 8154// CHECK-NEXT: [[TMP58:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 1155// CHECK-NEXT: store ptr [[TMP47]], ptr [[TMP58]], align 8156// CHECK-NEXT: [[TMP59:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 1157// CHECK-NEXT: store ptr [[ARRAYIDX4]], ptr [[TMP59]], align 8158// CHECK-NEXT: [[TMP60:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES11]], i32 0, i32 1159// CHECK-NEXT: store i64 [[TMP50]], ptr [[TMP60]], align 8160// CHECK-NEXT: [[TMP61:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i64 0, i64 1161// CHECK-NEXT: store ptr null, ptr [[TMP61]], align 8162// CHECK-NEXT: [[TMP62:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 2163// CHECK-NEXT: store ptr [[TMP51]], ptr [[TMP62]], align 8164// CHECK-NEXT: [[TMP63:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 2165// CHECK-NEXT: store ptr [[ARRAYIDX6]], ptr [[TMP63]], align 8166// CHECK-NEXT: [[TMP64:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES11]], i32 0, i32 2167// CHECK-NEXT: store i64 [[TMP54]], ptr [[TMP64]], align 8168// CHECK-NEXT: [[TMP65:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i64 0, i64 2169// CHECK-NEXT: store ptr null, ptr [[TMP65]], align 8170// CHECK-NEXT: [[TMP66:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0171// CHECK-NEXT: [[TMP67:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0172// CHECK-NEXT: [[TMP68:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES11]], i32 0, i32 0173// CHECK-NEXT: [[TMP69:%.*]] = load i32, ptr [[N_ADDR]], align 4174// CHECK-NEXT: store i32 [[TMP69]], ptr [[DOTCAPTURE_EXPR_]], align 4175// CHECK-NEXT: [[TMP70:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4176// CHECK-NEXT: [[SUB:%.*]] = sub i32 [[TMP70]], -2177// CHECK-NEXT: [[DIV:%.*]] = udiv i32 [[SUB]], 3178// CHECK-NEXT: [[SUB13:%.*]] = sub i32 [[DIV]], 1179// CHECK-NEXT: store i32 [[SUB13]], ptr [[DOTCAPTURE_EXPR_12]], align 4180// CHECK-NEXT: [[TMP71:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_12]], align 4181// CHECK-NEXT: [[ADD:%.*]] = add i32 [[TMP71]], 1182// CHECK-NEXT: [[TMP72:%.*]] = zext i32 [[ADD]] to i64183// CHECK-NEXT: [[TMP73:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS14]], i32 0, i32 0184// CHECK-NEXT: store i32 3, ptr [[TMP73]], align 4185// CHECK-NEXT: [[TMP74:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS14]], i32 0, i32 1186// CHECK-NEXT: store i32 3, ptr [[TMP74]], align 4187// CHECK-NEXT: [[TMP75:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS14]], i32 0, i32 2188// CHECK-NEXT: store ptr [[TMP66]], ptr [[TMP75]], align 8189// CHECK-NEXT: [[TMP76:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS14]], i32 0, i32 3190// CHECK-NEXT: store ptr [[TMP67]], ptr [[TMP76]], align 8191// CHECK-NEXT: [[TMP77:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS14]], i32 0, i32 4192// CHECK-NEXT: store ptr [[TMP68]], ptr [[TMP77]], align 8193// CHECK-NEXT: [[TMP78:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS14]], i32 0, i32 5194// CHECK-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP78]], align 8195// CHECK-NEXT: [[TMP79:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS14]], i32 0, i32 6196// CHECK-NEXT: store ptr null, ptr [[TMP79]], align 8197// CHECK-NEXT: [[TMP80:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS14]], i32 0, i32 7198// CHECK-NEXT: store ptr null, ptr [[TMP80]], align 8199// CHECK-NEXT: [[TMP81:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS14]], i32 0, i32 8200// CHECK-NEXT: store i64 [[TMP72]], ptr [[TMP81]], align 8201// CHECK-NEXT: [[TMP82:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS14]], i32 0, i32 9202// CHECK-NEXT: store i64 0, ptr [[TMP82]], align 8203// CHECK-NEXT: [[TMP83:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS14]], i32 0, i32 10204// CHECK-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP83]], align 4205// CHECK-NEXT: [[TMP84:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS14]], i32 0, i32 11206// CHECK-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP84]], align 4207// CHECK-NEXT: [[TMP85:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS14]], i32 0, i32 12208// CHECK-NEXT: store i32 0, ptr [[TMP85]], align 4209// CHECK-NEXT: [[TMP86:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18kernel_within_loopPiS_ii_l13.region_id, ptr [[KERNEL_ARGS14]])210// CHECK-NEXT: [[TMP87:%.*]] = icmp ne i32 [[TMP86]], 0211// CHECK-NEXT: br i1 [[TMP87]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]]212// CHECK: omp_offload.failed15:213// CHECK-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18kernel_within_loopPiS_ii_l13(i64 [[TMP44]], ptr [[TMP45]], ptr [[TMP46]]) #[[ATTR2]]214// CHECK-NEXT: br label [[OMP_OFFLOAD_CONT16]]215// CHECK: omp_offload.cont16:216// CHECK-NEXT: br label [[FOR_INC:%.*]]217// CHECK: for.inc:218// CHECK-NEXT: [[TMP88:%.*]] = load i32, ptr [[I]], align 4219// CHECK-NEXT: [[INC:%.*]] = add nsw i32 [[TMP88]], 1220// CHECK-NEXT: store i32 [[INC]], ptr [[I]], align 4221// CHECK-NEXT: br label [[FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]]222// CHECK: for.end:223// CHECK-NEXT: [[TMP89:%.*]] = load ptr, ptr [[A_ADDR]], align 8224// CHECK-NEXT: [[TMP90:%.*]] = load i32, ptr [[N_ADDR]], align 4225// CHECK-NEXT: [[SUB17:%.*]] = sub nsw i32 [[TMP90]], 1226// CHECK-NEXT: [[IDXPROM:%.*]] = sext i32 [[SUB17]] to i64227// CHECK-NEXT: [[ARRAYIDX18:%.*]] = getelementptr inbounds i32, ptr [[TMP89]], i64 [[IDXPROM]]228// CHECK-NEXT: [[TMP91:%.*]] = load i32, ptr [[ARRAYIDX18]], align 4229// CHECK-NEXT: ret i32 [[TMP91]]230//231//232// CHECK-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18kernel_within_loopPiS_ii_l9233// CHECK-SAME: (i64 noundef [[N:%.*]], ptr noundef [[A:%.*]], ptr noundef [[B:%.*]]) #[[ATTR1:[0-9]+]] {234// CHECK-NEXT: entry:235// CHECK-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8236// CHECK-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8237// CHECK-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8238// CHECK-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8239// CHECK-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8240// CHECK-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8241// CHECK-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8242// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[N_ADDR]], align 4243// CHECK-NEXT: store i32 [[TMP0]], ptr [[N_CASTED]], align 4244// CHECK-NEXT: [[TMP1:%.*]] = load i64, ptr [[N_CASTED]], align 8245// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8246// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[B_ADDR]], align 8247// CHECK-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB2]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18kernel_within_loopPiS_ii_l9.omp_outlined, i64 [[TMP1]], ptr [[TMP2]], ptr [[TMP3]])248// CHECK-NEXT: ret void249//250//251// CHECK-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18kernel_within_loopPiS_ii_l9.omp_outlined252// CHECK-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef [[A:%.*]], ptr noundef [[B:%.*]]) #[[ATTR1]] {253// CHECK-NEXT: entry:254// CHECK-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8255// CHECK-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8256// CHECK-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8257// CHECK-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8258// CHECK-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8259// CHECK-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4260// CHECK-NEXT: [[TMP:%.*]] = alloca i32, align 4261// CHECK-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4262// CHECK-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4263// CHECK-NEXT: [[J:%.*]] = alloca i32, align 4264// CHECK-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4265// CHECK-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4266// CHECK-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4267// CHECK-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4268// CHECK-NEXT: [[J3:%.*]] = alloca i32, align 4269// CHECK-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8270// CHECK-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8271// CHECK-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8272// CHECK-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8273// CHECK-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8274// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[N_ADDR]], align 4275// CHECK-NEXT: store i32 [[TMP0]], ptr [[DOTCAPTURE_EXPR_]], align 4276// CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4277// CHECK-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP1]], 0278// CHECK-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1279// CHECK-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1280// CHECK-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4281// CHECK-NEXT: store i32 0, ptr [[J]], align 4282// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4283// CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP2]]284// CHECK-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]285// CHECK: omp.precond.then:286// CHECK-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4287// CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4288// CHECK-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_UB]], align 4289// CHECK-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4290// CHECK-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4291// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8292// CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4293// CHECK-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP5]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)294// CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4295// CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4296// CHECK-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP6]], [[TMP7]]297// CHECK-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]298// CHECK: cond.true:299// CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4300// CHECK-NEXT: br label [[COND_END:%.*]]301// CHECK: cond.false:302// CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4303// CHECK-NEXT: br label [[COND_END]]304// CHECK: cond.end:305// CHECK-NEXT: [[COND:%.*]] = phi i32 [ [[TMP8]], [[COND_TRUE]] ], [ [[TMP9]], [[COND_FALSE]] ]306// CHECK-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4307// CHECK-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4308// CHECK-NEXT: store i32 [[TMP10]], ptr [[DOTOMP_IV]], align 4309// CHECK-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]310// CHECK: omp.inner.for.cond:311// CHECK-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4312// CHECK-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4313// CHECK-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP11]], [[TMP12]]314// CHECK-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]315// CHECK: omp.inner.for.body:316// CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4317// CHECK-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP13]], 1318// CHECK-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]]319// CHECK-NEXT: store i32 [[ADD]], ptr [[J3]], align 4320// CHECK-NEXT: [[TMP14:%.*]] = load ptr, ptr [[B_ADDR]], align 8321// CHECK-NEXT: [[TMP15:%.*]] = load i32, ptr [[J3]], align 4322// CHECK-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP15]] to i64323// CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP14]], i64 [[IDXPROM]]324// CHECK-NEXT: [[TMP16:%.*]] = load i32, ptr [[ARRAYIDX]], align 4325// CHECK-NEXT: [[TMP17:%.*]] = load ptr, ptr [[A_ADDR]], align 8326// CHECK-NEXT: [[TMP18:%.*]] = load i32, ptr [[J3]], align 4327// CHECK-NEXT: [[IDXPROM6:%.*]] = sext i32 [[TMP18]] to i64328// CHECK-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds i32, ptr [[TMP17]], i64 [[IDXPROM6]]329// CHECK-NEXT: store i32 [[TMP16]], ptr [[ARRAYIDX7]], align 4330// CHECK-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]331// CHECK: omp.body.continue:332// CHECK-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]333// CHECK: omp.inner.for.inc:334// CHECK-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4335// CHECK-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP19]], 1336// CHECK-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4337// CHECK-NEXT: br label [[OMP_INNER_FOR_COND]]338// CHECK: omp.inner.for.end:339// CHECK-NEXT: br label [[OMP_LOOP_EXIT:%.*]]340// CHECK: omp.loop.exit:341// CHECK-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8342// CHECK-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4343// CHECK-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP21]])344// CHECK-NEXT: br label [[OMP_PRECOND_END]]345// CHECK: omp.precond.end:346// CHECK-NEXT: ret void347//348//349// CHECK-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18kernel_within_loopPiS_ii_l13350// CHECK-SAME: (i64 noundef [[N:%.*]], ptr noundef [[A:%.*]], ptr noundef [[B:%.*]]) #[[ATTR1]] {351// CHECK-NEXT: entry:352// CHECK-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8353// CHECK-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8354// CHECK-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8355// CHECK-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8356// CHECK-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8357// CHECK-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8358// CHECK-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8359// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[N_ADDR]], align 4360// CHECK-NEXT: store i32 [[TMP0]], ptr [[N_CASTED]], align 4361// CHECK-NEXT: [[TMP1:%.*]] = load i64, ptr [[N_CASTED]], align 8362// CHECK-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8363// CHECK-NEXT: [[TMP3:%.*]] = load ptr, ptr [[B_ADDR]], align 8364// CHECK-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18kernel_within_loopPiS_ii_l13.omp_outlined, i64 [[TMP1]], ptr [[TMP2]], ptr [[TMP3]])365// CHECK-NEXT: ret void366//367//368// CHECK-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18kernel_within_loopPiS_ii_l13.omp_outlined369// CHECK-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], ptr noundef [[A:%.*]], ptr noundef [[B:%.*]]) #[[ATTR1]] {370// CHECK-NEXT: entry:371// CHECK-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8372// CHECK-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8373// CHECK-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8374// CHECK-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8375// CHECK-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8376// CHECK-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4377// CHECK-NEXT: [[TMP:%.*]] = alloca i32, align 4378// CHECK-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4379// CHECK-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4380// CHECK-NEXT: [[J:%.*]] = alloca i32, align 4381// CHECK-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4382// CHECK-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4383// CHECK-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4384// CHECK-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4385// CHECK-NEXT: [[J3:%.*]] = alloca i32, align 4386// CHECK-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8387// CHECK-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8388// CHECK-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8389// CHECK-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8390// CHECK-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8391// CHECK-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8392// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[N_ADDR]], align 4393// CHECK-NEXT: store i32 [[TMP0]], ptr [[DOTCAPTURE_EXPR_]], align 4394// CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4395// CHECK-NEXT: [[SUB:%.*]] = sub i32 [[TMP1]], -2396// CHECK-NEXT: [[DIV:%.*]] = udiv i32 [[SUB]], 3397// CHECK-NEXT: [[SUB2:%.*]] = sub i32 [[DIV]], 1398// CHECK-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4399// CHECK-NEXT: store i32 0, ptr [[J]], align 4400// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4401// CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP2]]402// CHECK-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]403// CHECK: omp.precond.then:404// CHECK-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4405// CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4406// CHECK-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_COMB_UB]], align 4407// CHECK-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4408// CHECK-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4409// CHECK-NEXT: [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8410// CHECK-NEXT: [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4411// CHECK-NEXT: call void @__kmpc_for_static_init_4u(ptr @[[GLOB3:[0-9]+]], i32 [[TMP5]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)412// CHECK-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4413// CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4414// CHECK-NEXT: [[CMP4:%.*]] = icmp ugt i32 [[TMP6]], [[TMP7]]415// CHECK-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]416// CHECK: cond.true:417// CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4418// CHECK-NEXT: br label [[COND_END:%.*]]419// CHECK: cond.false:420// CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4421// CHECK-NEXT: br label [[COND_END]]422// CHECK: cond.end:423// CHECK-NEXT: [[COND:%.*]] = phi i32 [ [[TMP8]], [[COND_TRUE]] ], [ [[TMP9]], [[COND_FALSE]] ]424// CHECK-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4425// CHECK-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4426// CHECK-NEXT: store i32 [[TMP10]], ptr [[DOTOMP_IV]], align 4427// CHECK-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]428// CHECK: omp.inner.for.cond:429// CHECK-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4430// CHECK-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4431// CHECK-NEXT: [[ADD:%.*]] = add i32 [[TMP12]], 1432// CHECK-NEXT: [[CMP5:%.*]] = icmp ult i32 [[TMP11]], [[ADD]]433// CHECK-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]434// CHECK: omp.inner.for.body:435// CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4436// CHECK-NEXT: [[TMP14:%.*]] = zext i32 [[TMP13]] to i64437// CHECK-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4438// CHECK-NEXT: [[TMP16:%.*]] = zext i32 [[TMP15]] to i64439// CHECK-NEXT: [[TMP17:%.*]] = load i32, ptr [[N_ADDR]], align 4440// CHECK-NEXT: store i32 [[TMP17]], ptr [[N_CASTED]], align 4441// CHECK-NEXT: [[TMP18:%.*]] = load i64, ptr [[N_CASTED]], align 8442// CHECK-NEXT: [[TMP19:%.*]] = load ptr, ptr [[A_ADDR]], align 8443// CHECK-NEXT: [[TMP20:%.*]] = load ptr, ptr [[B_ADDR]], align 8444// CHECK-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB2]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18kernel_within_loopPiS_ii_l13.omp_outlined.omp_outlined, i64 [[TMP14]], i64 [[TMP16]], i64 [[TMP18]], ptr [[TMP19]], ptr [[TMP20]])445// CHECK-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]446// CHECK: omp.inner.for.inc:447// CHECK-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4448// CHECK-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4449// CHECK-NEXT: [[ADD6:%.*]] = add i32 [[TMP21]], [[TMP22]]450// CHECK-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4451// CHECK-NEXT: br label [[OMP_INNER_FOR_COND]]452// CHECK: omp.inner.for.end:453// CHECK-NEXT: br label [[OMP_LOOP_EXIT:%.*]]454// CHECK: omp.loop.exit:455// CHECK-NEXT: [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8456// CHECK-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4457// CHECK-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB3]], i32 [[TMP24]])458// CHECK-NEXT: br label [[OMP_PRECOND_END]]459// CHECK: omp.precond.end:460// CHECK-NEXT: ret void461//462//463// CHECK-LABEL: define internal void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z18kernel_within_loopPiS_ii_l13.omp_outlined.omp_outlined464// CHECK-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i64 noundef [[N:%.*]], ptr noundef [[A:%.*]], ptr noundef [[B:%.*]]) #[[ATTR1]] {465// CHECK-NEXT: entry:466// CHECK-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8467// CHECK-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8468// CHECK-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8469// CHECK-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8470// CHECK-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8471// CHECK-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8472// CHECK-NEXT: [[B_ADDR:%.*]] = alloca ptr, align 8473// CHECK-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4474// CHECK-NEXT: [[TMP:%.*]] = alloca i32, align 4475// CHECK-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4476// CHECK-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4477// CHECK-NEXT: [[J:%.*]] = alloca i32, align 4478// CHECK-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4479// CHECK-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4480// CHECK-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4481// CHECK-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4482// CHECK-NEXT: [[J4:%.*]] = alloca i32, align 4483// CHECK-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8484// CHECK-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8485// CHECK-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8486// CHECK-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8487// CHECK-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8488// CHECK-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8489// CHECK-NEXT: store ptr [[B]], ptr [[B_ADDR]], align 8490// CHECK-NEXT: [[TMP0:%.*]] = load i32, ptr [[N_ADDR]], align 4491// CHECK-NEXT: store i32 [[TMP0]], ptr [[DOTCAPTURE_EXPR_]], align 4492// CHECK-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4493// CHECK-NEXT: [[SUB:%.*]] = sub i32 [[TMP1]], -2494// CHECK-NEXT: [[DIV:%.*]] = udiv i32 [[SUB]], 3495// CHECK-NEXT: [[SUB2:%.*]] = sub i32 [[DIV]], 1496// CHECK-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4497// CHECK-NEXT: store i32 0, ptr [[J]], align 4498// CHECK-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4499// CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP2]]500// CHECK-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]]501// CHECK: omp.precond.then:502// CHECK-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4503// CHECK-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4504// CHECK-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_UB]], align 4505// CHECK-NEXT: [[TMP4:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8506// CHECK-NEXT: [[CONV:%.*]] = trunc i64 [[TMP4]] to i32507// CHECK-NEXT: [[TMP5:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8508// CHECK-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP5]] to i32509// CHECK-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4510// CHECK-NEXT: store i32 [[CONV3]], ptr [[DOTOMP_UB]], align 4511// CHECK-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4512// CHECK-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4513// CHECK-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8514// CHECK-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4515// CHECK-NEXT: call void @__kmpc_for_static_init_4u(ptr @[[GLOB1]], i32 [[TMP7]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)516// CHECK-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4517// CHECK-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4518// CHECK-NEXT: [[CMP5:%.*]] = icmp ugt i32 [[TMP8]], [[TMP9]]519// CHECK-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]520// CHECK: cond.true:521// CHECK-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4522// CHECK-NEXT: br label [[COND_END:%.*]]523// CHECK: cond.false:524// CHECK-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4525// CHECK-NEXT: br label [[COND_END]]526// CHECK: cond.end:527// CHECK-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ]528// CHECK-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4529// CHECK-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4530// CHECK-NEXT: store i32 [[TMP12]], ptr [[DOTOMP_IV]], align 4531// CHECK-NEXT: br label [[OMP_INNER_FOR_COND:%.*]]532// CHECK: omp.inner.for.cond:533// CHECK-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4534// CHECK-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4535// CHECK-NEXT: [[ADD:%.*]] = add i32 [[TMP14]], 1536// CHECK-NEXT: [[CMP6:%.*]] = icmp ult i32 [[TMP13]], [[ADD]]537// CHECK-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]538// CHECK: omp.inner.for.body:539// CHECK-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4540// CHECK-NEXT: [[MUL:%.*]] = mul i32 [[TMP15]], 3541// CHECK-NEXT: [[ADD7:%.*]] = add i32 0, [[MUL]]542// CHECK-NEXT: store i32 [[ADD7]], ptr [[J4]], align 4543// CHECK-NEXT: [[TMP16:%.*]] = load ptr, ptr [[B_ADDR]], align 8544// CHECK-NEXT: [[TMP17:%.*]] = load i32, ptr [[J4]], align 4545// CHECK-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP17]] to i64546// CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP16]], i64 [[IDXPROM]]547// CHECK-NEXT: [[TMP18:%.*]] = load i32, ptr [[ARRAYIDX]], align 4548// CHECK-NEXT: [[MUL8:%.*]] = mul nsw i32 [[TMP18]], 2549// CHECK-NEXT: [[TMP19:%.*]] = load ptr, ptr [[A_ADDR]], align 8550// CHECK-NEXT: [[TMP20:%.*]] = load i32, ptr [[J4]], align 4551// CHECK-NEXT: [[IDXPROM9:%.*]] = sext i32 [[TMP20]] to i64552// CHECK-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i32, ptr [[TMP19]], i64 [[IDXPROM9]]553// CHECK-NEXT: store i32 [[MUL8]], ptr [[ARRAYIDX10]], align 4554// CHECK-NEXT: br label [[OMP_BODY_CONTINUE:%.*]]555// CHECK: omp.body.continue:556// CHECK-NEXT: br label [[OMP_INNER_FOR_INC:%.*]]557// CHECK: omp.inner.for.inc:558// CHECK-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4559// CHECK-NEXT: [[ADD11:%.*]] = add i32 [[TMP21]], 1560// CHECK-NEXT: store i32 [[ADD11]], ptr [[DOTOMP_IV]], align 4561// CHECK-NEXT: br label [[OMP_INNER_FOR_COND]]562// CHECK: omp.inner.for.end:563// CHECK-NEXT: br label [[OMP_LOOP_EXIT:%.*]]564// CHECK: omp.loop.exit:565// CHECK-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8566// CHECK-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4567// CHECK-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP23]])568// CHECK-NEXT: br label [[OMP_PRECOND_END]]569// CHECK: omp.precond.end:570// CHECK-NEXT: ret void571//572