1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _ 2 // RUN: %clang_cc1 -verify -fopenmp -x c++ -triple x86_64-unknown-linux -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s --check-prefix=CHECK1 3 // RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-unknown-linux -fexceptions -fcxx-exceptions -emit-pch -o %t %s 4 // RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-unknown-linux -fexceptions -fcxx-exceptions -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK2 5 6 // RUN: %clang_cc1 -triple x86_64-unknown-linux -verify -fopenmp-simd -x c++ -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 7 // RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s 8 // RUN: %clang_cc1 -fopenmp-simd -x c++ -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -debug-info-kind=limited -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 9 // expected-no-diagnostics 10 #ifndef HEADER 11 #define HEADER 12 13 int main(int argc, char **argv) { 14 #pragma omp parallel for reduction(task, +: argc, argv[0:10][0:argc]) 15 for (long long i = 0; i < 10; ++i) { 16 #pragma omp task in_reduction(+: argc, argv[0:10][0:argc]) 17 ; 18 } 19 } 20 21 22 23 // Init firstprivate copy of argc 24 25 // Init firstprivate copy of argv[0:10][0:argc] 26 27 // Register task reduction. 28 29 30 31 32 33 34 35 36 37 #endif 38 // CHECK1-LABEL: define {{[^@]+}}@main 39 // CHECK1-SAME: (i32 [[ARGC:%.*]], i8** [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 40 // CHECK1-NEXT: entry: 41 // CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 42 // CHECK1-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 43 // CHECK1-NEXT: [[OMP_OUTLINED_ARG_AGG_:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8 44 // CHECK1-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 45 // CHECK1-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 46 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[OMP_OUTLINED_ARG_AGG_]], i32 0, i32 0 47 // CHECK1-NEXT: store i32* [[ARGC_ADDR]], i32** [[TMP0]], align 8 48 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[OMP_OUTLINED_ARG_AGG_]], i32 0, i32 1 49 // CHECK1-NEXT: [[TMP2:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 50 // CHECK1-NEXT: store i8** [[TMP2]], i8*** [[TMP1]], align 8 51 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.anon*)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.anon* [[OMP_OUTLINED_ARG_AGG_]]) 52 // CHECK1-NEXT: ret i32 0 53 // 54 // 55 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined. 56 // CHECK1-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.anon* noalias [[__CONTEXT:%.*]]) #[[ATTR1:[0-9]+]] { 57 // CHECK1-NEXT: entry: 58 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 59 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 60 // CHECK1-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca %struct.anon*, align 8 61 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 62 // CHECK1-NEXT: [[TMP:%.*]] = alloca i64, align 8 63 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8 64 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8 65 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8 66 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 67 // CHECK1-NEXT: [[ARGC:%.*]] = alloca i32, align 4 68 // CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 69 // CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 70 // CHECK1-NEXT: [[_TMP4:%.*]] = alloca i8**, align 8 71 // CHECK1-NEXT: [[_TMP5:%.*]] = alloca i8*, align 8 72 // CHECK1-NEXT: [[DOTRD_INPUT_:%.*]] = alloca [2 x %struct.kmp_taskred_input_t], align 8 73 // CHECK1-NEXT: [[DOTTASK_RED_:%.*]] = alloca i8*, align 8 74 // CHECK1-NEXT: [[I:%.*]] = alloca i64, align 8 75 // CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_0:%.*]], align 8 76 // CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8 77 // CHECK1-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1 78 // CHECK1-NEXT: [[_TMP27:%.*]] = alloca i8, align 1 79 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 80 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 81 // CHECK1-NEXT: store %struct.anon* [[__CONTEXT]], %struct.anon** [[__CONTEXT_ADDR]], align 8 82 // CHECK1-NEXT: [[TMP0:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR]], align 8 83 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP0]], i32 0, i32 0 84 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[TMP1]], align 8 85 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP0]], i32 0, i32 1 86 // CHECK1-NEXT: store i64 0, i64* [[DOTOMP_LB]], align 8 87 // CHECK1-NEXT: store i64 9, i64* [[DOTOMP_UB]], align 8 88 // CHECK1-NEXT: store i64 1, i64* [[DOTOMP_STRIDE]], align 8 89 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 90 // CHECK1-NEXT: store i32 0, i32* [[ARGC]], align 4 91 // CHECK1-NEXT: [[TMP4:%.*]] = load i8**, i8*** [[TMP3]], align 8 92 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8*, i8** [[TMP4]], i64 0 93 // CHECK1-NEXT: [[TMP5:%.*]] = load i8*, i8** [[ARRAYIDX]], align 8 94 // CHECK1-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, i8* [[TMP5]], i64 0 95 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP2]], align 4 96 // CHECK1-NEXT: [[TMP7:%.*]] = sext i32 [[TMP6]] to i64 97 // CHECK1-NEXT: [[LB_ADD_LEN:%.*]] = add nsw i64 -1, [[TMP7]] 98 // CHECK1-NEXT: [[TMP8:%.*]] = load i8**, i8*** [[TMP3]], align 8 99 // CHECK1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8*, i8** [[TMP8]], i64 9 100 // CHECK1-NEXT: [[TMP9:%.*]] = load i8*, i8** [[ARRAYIDX2]], align 8 101 // CHECK1-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds i8, i8* [[TMP9]], i64 [[LB_ADD_LEN]] 102 // CHECK1-NEXT: [[TMP10:%.*]] = ptrtoint i8* [[ARRAYIDX3]] to i64 103 // CHECK1-NEXT: [[TMP11:%.*]] = ptrtoint i8* [[ARRAYIDX1]] to i64 104 // CHECK1-NEXT: [[TMP12:%.*]] = sub i64 [[TMP10]], [[TMP11]] 105 // CHECK1-NEXT: [[TMP13:%.*]] = sdiv exact i64 [[TMP12]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 106 // CHECK1-NEXT: [[TMP14:%.*]] = add nuw i64 [[TMP13]], 1 107 // CHECK1-NEXT: [[TMP15:%.*]] = mul nuw i64 [[TMP14]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 108 // CHECK1-NEXT: [[TMP16:%.*]] = call i8* @llvm.stacksave() 109 // CHECK1-NEXT: store i8* [[TMP16]], i8** [[SAVED_STACK]], align 8 110 // CHECK1-NEXT: [[VLA:%.*]] = alloca i8, i64 [[TMP14]], align 16 111 // CHECK1-NEXT: store i64 [[TMP14]], i64* [[__VLA_EXPR0]], align 8 112 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP14]] 113 // CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[VLA]], [[TMP17]] 114 // CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 115 // CHECK1: omp.arrayinit.body: 116 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 117 // CHECK1-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 118 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 119 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP17]] 120 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 121 // CHECK1: omp.arrayinit.done: 122 // CHECK1-NEXT: [[TMP18:%.*]] = load i8**, i8*** [[TMP3]], align 8 123 // CHECK1-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8 124 // CHECK1-NEXT: [[TMP20:%.*]] = ptrtoint i8* [[TMP19]] to i64 125 // CHECK1-NEXT: [[TMP21:%.*]] = ptrtoint i8* [[ARRAYIDX1]] to i64 126 // CHECK1-NEXT: [[TMP22:%.*]] = sub i64 [[TMP20]], [[TMP21]] 127 // CHECK1-NEXT: [[TMP23:%.*]] = sdiv exact i64 [[TMP22]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 128 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP23]] 129 // CHECK1-NEXT: store i8** [[_TMP5]], i8*** [[_TMP4]], align 8 130 // CHECK1-NEXT: store i8* [[TMP24]], i8** [[_TMP5]], align 8 131 // CHECK1-NEXT: [[DOTRD_INPUT_GEP_:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]], i64 0, i64 0 132 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T:%.*]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 0 133 // CHECK1-NEXT: [[TMP26:%.*]] = bitcast i32* [[ARGC]] to i8* 134 // CHECK1-NEXT: store i8* [[TMP26]], i8** [[TMP25]], align 8 135 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 1 136 // CHECK1-NEXT: [[TMP28:%.*]] = bitcast i32* [[TMP2]] to i8* 137 // CHECK1-NEXT: store i8* [[TMP28]], i8** [[TMP27]], align 8 138 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 2 139 // CHECK1-NEXT: store i64 4, i64* [[TMP29]], align 8 140 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 3 141 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init. to i8*), i8** [[TMP30]], align 8 142 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 4 143 // CHECK1-NEXT: store i8* null, i8** [[TMP31]], align 8 144 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 5 145 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb. to i8*), i8** [[TMP32]], align 8 146 // CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 6 147 // CHECK1-NEXT: [[TMP34:%.*]] = bitcast i32* [[TMP33]] to i8* 148 // CHECK1-NEXT: call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP34]], i8 0, i64 4, i1 false) 149 // CHECK1-NEXT: [[DOTRD_INPUT_GEP_6:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]], i64 0, i64 1 150 // CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 0 151 // CHECK1-NEXT: [[TMP36:%.*]] = load i8**, i8*** [[TMP3]], align 8 152 // CHECK1-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds i8*, i8** [[TMP36]], i64 0 153 // CHECK1-NEXT: [[TMP37:%.*]] = load i8*, i8** [[ARRAYIDX7]], align 8 154 // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i8, i8* [[TMP37]], i64 0 155 // CHECK1-NEXT: [[TMP38:%.*]] = load i32, i32* [[TMP2]], align 4 156 // CHECK1-NEXT: [[TMP39:%.*]] = sext i32 [[TMP38]] to i64 157 // CHECK1-NEXT: [[LB_ADD_LEN9:%.*]] = add nsw i64 -1, [[TMP39]] 158 // CHECK1-NEXT: [[TMP40:%.*]] = load i8**, i8*** [[TMP3]], align 8 159 // CHECK1-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i8*, i8** [[TMP40]], i64 9 160 // CHECK1-NEXT: [[TMP41:%.*]] = load i8*, i8** [[ARRAYIDX10]], align 8 161 // CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds i8, i8* [[TMP41]], i64 [[LB_ADD_LEN9]] 162 // CHECK1-NEXT: store i8* [[VLA]], i8** [[TMP35]], align 8 163 // CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 1 164 // CHECK1-NEXT: store i8* [[ARRAYIDX8]], i8** [[TMP42]], align 8 165 // CHECK1-NEXT: [[TMP43:%.*]] = ptrtoint i8* [[ARRAYIDX11]] to i64 166 // CHECK1-NEXT: [[TMP44:%.*]] = ptrtoint i8* [[ARRAYIDX8]] to i64 167 // CHECK1-NEXT: [[TMP45:%.*]] = sub i64 [[TMP43]], [[TMP44]] 168 // CHECK1-NEXT: [[TMP46:%.*]] = sdiv exact i64 [[TMP45]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 169 // CHECK1-NEXT: [[TMP47:%.*]] = add nuw i64 [[TMP46]], 1 170 // CHECK1-NEXT: [[TMP48:%.*]] = mul nuw i64 [[TMP47]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 171 // CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 2 172 // CHECK1-NEXT: store i64 [[TMP48]], i64* [[TMP49]], align 8 173 // CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 3 174 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init..1 to i8*), i8** [[TMP50]], align 8 175 // CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 4 176 // CHECK1-NEXT: store i8* null, i8** [[TMP51]], align 8 177 // CHECK1-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 5 178 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb..2 to i8*), i8** [[TMP52]], align 8 179 // CHECK1-NEXT: [[TMP53:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 6 180 // CHECK1-NEXT: store i32 1, i32* [[TMP53]], align 8 181 // CHECK1-NEXT: [[TMP54:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 182 // CHECK1-NEXT: [[TMP55:%.*]] = load i32, i32* [[TMP54]], align 4 183 // CHECK1-NEXT: [[TMP56:%.*]] = bitcast [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]] to i8* 184 // CHECK1-NEXT: [[TMP57:%.*]] = call i8* @__kmpc_taskred_modifier_init(%struct.ident_t* @[[GLOB1]], i32 [[TMP55]], i32 1, i32 2, i8* [[TMP56]]) 185 // CHECK1-NEXT: store i8* [[TMP57]], i8** [[DOTTASK_RED_]], align 8 186 // CHECK1-NEXT: [[TMP58:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 187 // CHECK1-NEXT: [[TMP59:%.*]] = load i32, i32* [[TMP58]], align 4 188 // CHECK1-NEXT: call void @__kmpc_for_static_init_8(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP59]], i32 34, i32* [[DOTOMP_IS_LAST]], i64* [[DOTOMP_LB]], i64* [[DOTOMP_UB]], i64* [[DOTOMP_STRIDE]], i64 1, i64 1) 189 // CHECK1-NEXT: [[TMP60:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 190 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i64 [[TMP60]], 9 191 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 192 // CHECK1: cond.true: 193 // CHECK1-NEXT: br label [[COND_END:%.*]] 194 // CHECK1: cond.false: 195 // CHECK1-NEXT: [[TMP61:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 196 // CHECK1-NEXT: br label [[COND_END]] 197 // CHECK1: cond.end: 198 // CHECK1-NEXT: [[COND:%.*]] = phi i64 [ 9, [[COND_TRUE]] ], [ [[TMP61]], [[COND_FALSE]] ] 199 // CHECK1-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8 200 // CHECK1-NEXT: [[TMP62:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8 201 // CHECK1-NEXT: store i64 [[TMP62]], i64* [[DOTOMP_IV]], align 8 202 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 203 // CHECK1: omp.inner.for.cond: 204 // CHECK1-NEXT: [[TMP63:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 205 // CHECK1-NEXT: [[TMP64:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 206 // CHECK1-NEXT: [[CMP12:%.*]] = icmp sle i64 [[TMP63]], [[TMP64]] 207 // CHECK1-NEXT: br i1 [[CMP12]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]] 208 // CHECK1: omp.inner.for.cond.cleanup: 209 // CHECK1-NEXT: br label [[OMP_INNER_FOR_END:%.*]] 210 // CHECK1: omp.inner.for.body: 211 // CHECK1-NEXT: [[TMP65:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 212 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i64 [[TMP65]], 1 213 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL]] 214 // CHECK1-NEXT: store i64 [[ADD]], i64* [[I]], align 8 215 // CHECK1-NEXT: [[TMP66:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[AGG_CAPTURED]], i32 0, i32 0 216 // CHECK1-NEXT: store i8** [[DOTTASK_RED_]], i8*** [[TMP66]], align 8 217 // CHECK1-NEXT: [[TMP67:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[AGG_CAPTURED]], i32 0, i32 1 218 // CHECK1-NEXT: store i32* [[ARGC]], i32** [[TMP67]], align 8 219 // CHECK1-NEXT: [[TMP68:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[AGG_CAPTURED]], i32 0, i32 2 220 // CHECK1-NEXT: [[TMP69:%.*]] = load i8**, i8*** [[_TMP4]], align 8 221 // CHECK1-NEXT: store i8** [[TMP69]], i8*** [[TMP68]], align 8 222 // CHECK1-NEXT: [[TMP70:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 223 // CHECK1-NEXT: [[TMP71:%.*]] = load i32, i32* [[TMP70]], align 4 224 // CHECK1-NEXT: [[TMP72:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP71]], i32 1, i64 48, i64 24, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @.omp_task_entry. to i32 (i32, i8*)*)) 225 // CHECK1-NEXT: [[TMP73:%.*]] = bitcast i8* [[TMP72]] to %struct.kmp_task_t_with_privates* 226 // CHECK1-NEXT: [[TMP74:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP73]], i32 0, i32 0 227 // CHECK1-NEXT: [[TMP75:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP74]], i32 0, i32 0 228 // CHECK1-NEXT: [[TMP76:%.*]] = load i8*, i8** [[TMP75]], align 8 229 // CHECK1-NEXT: [[TMP77:%.*]] = bitcast %struct.anon.0* [[AGG_CAPTURED]] to i8* 230 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP76]], i8* align 8 [[TMP77]], i64 24, i1 false) 231 // CHECK1-NEXT: [[TMP78:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP73]], i32 0, i32 1 232 // CHECK1-NEXT: [[TMP79:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP78]], i32 0, i32 0 233 // CHECK1-NEXT: [[TMP80:%.*]] = load i8*, i8** [[DOTTASK_RED_]], align 8 234 // CHECK1-NEXT: store i8* [[TMP80]], i8** [[TMP79]], align 8 235 // CHECK1-NEXT: [[TMP81:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 236 // CHECK1-NEXT: [[TMP82:%.*]] = load i32, i32* [[TMP81]], align 4 237 // CHECK1-NEXT: [[TMP83:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP82]], i8* [[TMP72]]) 238 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 239 // CHECK1: omp.body.continue: 240 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 241 // CHECK1: omp.inner.for.inc: 242 // CHECK1-NEXT: [[TMP84:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 243 // CHECK1-NEXT: [[ADD13:%.*]] = add nsw i64 [[TMP84]], 1 244 // CHECK1-NEXT: store i64 [[ADD13]], i64* [[DOTOMP_IV]], align 8 245 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 246 // CHECK1: omp.inner.for.end: 247 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 248 // CHECK1: omp.loop.exit: 249 // CHECK1-NEXT: [[TMP85:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 250 // CHECK1-NEXT: [[TMP86:%.*]] = load i32, i32* [[TMP85]], align 4 251 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP86]]) 252 // CHECK1-NEXT: [[TMP87:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 253 // CHECK1-NEXT: [[TMP88:%.*]] = load i32, i32* [[TMP87]], align 4 254 // CHECK1-NEXT: call void @__kmpc_task_reduction_modifier_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP88]], i32 1) 255 // CHECK1-NEXT: [[TMP89:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 256 // CHECK1-NEXT: [[TMP90:%.*]] = bitcast i32* [[ARGC]] to i8* 257 // CHECK1-NEXT: store i8* [[TMP90]], i8** [[TMP89]], align 8 258 // CHECK1-NEXT: [[TMP91:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 259 // CHECK1-NEXT: store i8* [[VLA]], i8** [[TMP91]], align 8 260 // CHECK1-NEXT: [[TMP92:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 261 // CHECK1-NEXT: [[TMP93:%.*]] = inttoptr i64 [[TMP14]] to i8* 262 // CHECK1-NEXT: store i8* [[TMP93]], i8** [[TMP92]], align 8 263 // CHECK1-NEXT: [[TMP94:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 264 // CHECK1-NEXT: [[TMP95:%.*]] = load i32, i32* [[TMP94]], align 4 265 // CHECK1-NEXT: [[TMP96:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 266 // CHECK1-NEXT: [[TMP97:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[TMP95]], i32 2, i64 24, i8* [[TMP96]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var) 267 // CHECK1-NEXT: switch i32 [[TMP97]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 268 // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 269 // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 270 // CHECK1-NEXT: ] 271 // CHECK1: .omp.reduction.case1: 272 // CHECK1-NEXT: [[TMP98:%.*]] = load i32, i32* [[TMP2]], align 4 273 // CHECK1-NEXT: [[TMP99:%.*]] = load i32, i32* [[ARGC]], align 4 274 // CHECK1-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP98]], [[TMP99]] 275 // CHECK1-NEXT: store i32 [[ADD14]], i32* [[TMP2]], align 4 276 // CHECK1-NEXT: [[TMP100:%.*]] = getelementptr i8, i8* [[ARRAYIDX1]], i64 [[TMP14]] 277 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[ARRAYIDX1]], [[TMP100]] 278 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE21:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 279 // CHECK1: omp.arraycpy.body: 280 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 281 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST15:%.*]] = phi i8* [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT19:%.*]], [[OMP_ARRAYCPY_BODY]] ] 282 // CHECK1-NEXT: [[TMP101:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST15]], align 1 283 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP101]] to i32 284 // CHECK1-NEXT: [[TMP102:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 285 // CHECK1-NEXT: [[CONV16:%.*]] = sext i8 [[TMP102]] to i32 286 // CHECK1-NEXT: [[ADD17:%.*]] = add nsw i32 [[CONV]], [[CONV16]] 287 // CHECK1-NEXT: [[CONV18:%.*]] = trunc i32 [[ADD17]] to i8 288 // CHECK1-NEXT: store i8 [[CONV18]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST15]], align 1 289 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT19]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST15]], i32 1 290 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 291 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE20:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT19]], [[TMP100]] 292 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE20]], label [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_BODY]] 293 // CHECK1: omp.arraycpy.done21: 294 // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP95]], [8 x i32]* @.gomp_critical_user_.reduction.var) 295 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 296 // CHECK1: .omp.reduction.case2: 297 // CHECK1-NEXT: [[TMP103:%.*]] = load i32, i32* [[ARGC]], align 4 298 // CHECK1-NEXT: [[TMP104:%.*]] = atomicrmw add i32* [[TMP2]], i32 [[TMP103]] monotonic, align 4 299 // CHECK1-NEXT: [[TMP105:%.*]] = getelementptr i8, i8* [[ARRAYIDX1]], i64 [[TMP14]] 300 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY22:%.*]] = icmp eq i8* [[ARRAYIDX1]], [[TMP105]] 301 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY22]], label [[OMP_ARRAYCPY_DONE35:%.*]], label [[OMP_ARRAYCPY_BODY23:%.*]] 302 // CHECK1: omp.arraycpy.body23: 303 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST24:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT33:%.*]], [[ATOMIC_EXIT:%.*]] ] 304 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST25:%.*]] = phi i8* [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT32:%.*]], [[ATOMIC_EXIT]] ] 305 // CHECK1-NEXT: [[TMP106:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST24]], align 1 306 // CHECK1-NEXT: [[CONV26:%.*]] = sext i8 [[TMP106]] to i32 307 // CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST25]] monotonic, align 1 308 // CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]] 309 // CHECK1: atomic_cont: 310 // CHECK1-NEXT: [[TMP107:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY23]] ], [ [[TMP112:%.*]], [[ATOMIC_CONT]] ] 311 // CHECK1-NEXT: store i8 [[TMP107]], i8* [[_TMP27]], align 1 312 // CHECK1-NEXT: [[TMP108:%.*]] = load i8, i8* [[_TMP27]], align 1 313 // CHECK1-NEXT: [[CONV28:%.*]] = sext i8 [[TMP108]] to i32 314 // CHECK1-NEXT: [[TMP109:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST24]], align 1 315 // CHECK1-NEXT: [[CONV29:%.*]] = sext i8 [[TMP109]] to i32 316 // CHECK1-NEXT: [[ADD30:%.*]] = add nsw i32 [[CONV28]], [[CONV29]] 317 // CHECK1-NEXT: [[CONV31:%.*]] = trunc i32 [[ADD30]] to i8 318 // CHECK1-NEXT: store i8 [[CONV31]], i8* [[ATOMIC_TEMP]], align 1 319 // CHECK1-NEXT: [[TMP110:%.*]] = load i8, i8* [[ATOMIC_TEMP]], align 1 320 // CHECK1-NEXT: [[TMP111:%.*]] = cmpxchg i8* [[OMP_ARRAYCPY_DESTELEMENTPAST25]], i8 [[TMP107]], i8 [[TMP110]] monotonic monotonic, align 1 321 // CHECK1-NEXT: [[TMP112]] = extractvalue { i8, i1 } [[TMP111]], 0 322 // CHECK1-NEXT: [[TMP113:%.*]] = extractvalue { i8, i1 } [[TMP111]], 1 323 // CHECK1-NEXT: br i1 [[TMP113]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]] 324 // CHECK1: atomic_exit: 325 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT32]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST25]], i32 1 326 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT33]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST24]], i32 1 327 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE34:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT32]], [[TMP105]] 328 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE34]], label [[OMP_ARRAYCPY_DONE35]], label [[OMP_ARRAYCPY_BODY23]] 329 // CHECK1: omp.arraycpy.done35: 330 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 331 // CHECK1: .omp.reduction.default: 332 // CHECK1-NEXT: [[TMP114:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 333 // CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP114]]) 334 // CHECK1-NEXT: ret void 335 // 336 // 337 // CHECK1-LABEL: define {{[^@]+}}@.red_init. 338 // CHECK1-SAME: (i8* noalias [[TMP0:%.*]], i8* noalias [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { 339 // CHECK1-NEXT: entry: 340 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 341 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 342 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 343 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 344 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 345 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to i32* 346 // CHECK1-NEXT: store i32 0, i32* [[TMP3]], align 8 347 // CHECK1-NEXT: ret void 348 // 349 // 350 // CHECK1-LABEL: define {{[^@]+}}@.red_comb. 351 // CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 352 // CHECK1-NEXT: entry: 353 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 354 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 355 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 356 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 357 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 358 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to i32* 359 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 360 // CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to i32* 361 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP3]], align 8 362 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP5]], align 8 363 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP6]], [[TMP7]] 364 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP3]], align 8 365 // CHECK1-NEXT: ret void 366 // 367 // 368 // CHECK1-LABEL: define {{[^@]+}}@.red_init..1 369 // CHECK1-SAME: (i8* noalias [[TMP0:%.*]], i8* noalias [[TMP1:%.*]]) #[[ATTR3]] { 370 // CHECK1-NEXT: entry: 371 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 372 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 373 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 374 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 375 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 376 // CHECK1-NEXT: [[TMP3:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8 377 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr i8, i8* [[TMP2]], i64 [[TMP3]] 378 // CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[TMP2]], [[TMP4]] 379 // CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 380 // CHECK1: omp.arrayinit.body: 381 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP2]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 382 // CHECK1-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 383 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 384 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]] 385 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 386 // CHECK1: omp.arrayinit.done: 387 // CHECK1-NEXT: ret void 388 // 389 // 390 // CHECK1-LABEL: define {{[^@]+}}@.red_comb..2 391 // CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 392 // CHECK1-NEXT: entry: 393 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 394 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 395 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 396 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 397 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8 398 // CHECK1-NEXT: [[TMP3:%.*]] = load i8*, i8** [[DOTADDR]], align 8 399 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 400 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr i8, i8* [[TMP3]], i64 [[TMP2]] 401 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP3]], [[TMP5]] 402 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 403 // CHECK1: omp.arraycpy.body: 404 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP4]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 405 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP3]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 406 // CHECK1-NEXT: [[TMP6:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 407 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP6]] to i32 408 // CHECK1-NEXT: [[TMP7:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 409 // CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP7]] to i32 410 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 411 // CHECK1-NEXT: [[CONV3:%.*]] = trunc i32 [[ADD]] to i8 412 // CHECK1-NEXT: store i8 [[CONV3]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 413 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 414 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 415 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP5]] 416 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]] 417 // CHECK1: omp.arraycpy.done4: 418 // CHECK1-NEXT: ret void 419 // 420 // 421 // CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map. 422 // CHECK1-SAME: (%struct..kmp_privates.t* noalias [[TMP0:%.*]], i8*** noalias [[TMP1:%.*]]) #[[ATTR6:[0-9]+]] { 423 // CHECK1-NEXT: entry: 424 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca %struct..kmp_privates.t*, align 8 425 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8***, align 8 426 // CHECK1-NEXT: store %struct..kmp_privates.t* [[TMP0]], %struct..kmp_privates.t** [[DOTADDR]], align 8 427 // CHECK1-NEXT: store i8*** [[TMP1]], i8**** [[DOTADDR1]], align 8 428 // CHECK1-NEXT: [[TMP2:%.*]] = load %struct..kmp_privates.t*, %struct..kmp_privates.t** [[DOTADDR]], align 8 429 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP2]], i32 0, i32 0 430 // CHECK1-NEXT: [[TMP4:%.*]] = load i8***, i8**** [[DOTADDR1]], align 8 431 // CHECK1-NEXT: store i8** [[TMP3]], i8*** [[TMP4]], align 8 432 // CHECK1-NEXT: ret void 433 // 434 // 435 // CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry. 436 // CHECK1-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias [[TMP1:%.*]]) #[[ATTR3]] { 437 // CHECK1-NEXT: entry: 438 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4 439 // CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8 440 // CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8 441 // CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8 442 // CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8 443 // CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon.0*, align 8 444 // CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca i8**, align 8 445 // CHECK1-NEXT: [[TMP_I:%.*]] = alloca i8**, align 8 446 // CHECK1-NEXT: [[TMP4_I:%.*]] = alloca i8*, align 8 447 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4 448 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8 449 // CHECK1-NEXT: store i32 [[TMP0]], i32* [[DOTADDR]], align 4 450 // CHECK1-NEXT: store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8 451 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4 452 // CHECK1-NEXT: [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8 453 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0 454 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2 455 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0 456 // CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 457 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon.0* 458 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1 459 // CHECK1-NEXT: [[TMP10:%.*]] = bitcast %struct..kmp_privates.t* [[TMP9]] to i8* 460 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast %struct.kmp_task_t_with_privates* [[TMP3]] to i8* 461 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META3:![0-9]+]]) 462 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META6:![0-9]+]]) 463 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META8:![0-9]+]]) 464 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]]) 465 // CHECK1-NEXT: store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 466 // CHECK1-NEXT: store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !12 467 // CHECK1-NEXT: store i8* [[TMP10]], i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 468 // CHECK1-NEXT: store void (i8*, ...)* bitcast (void (%struct..kmp_privates.t*, i8***)* @.omp_task_privates_map. to void (i8*, ...)*), void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12 469 // CHECK1-NEXT: store i8* [[TMP11]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !12 470 // CHECK1-NEXT: store %struct.anon.0* [[TMP8]], %struct.anon.0** [[__CONTEXT_ADDR_I]], align 8, !noalias !12 471 // CHECK1-NEXT: [[TMP12:%.*]] = load %struct.anon.0*, %struct.anon.0** [[__CONTEXT_ADDR_I]], align 8, !noalias !12 472 // CHECK1-NEXT: [[TMP13:%.*]] = load void (i8*, ...)*, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12 473 // CHECK1-NEXT: [[TMP14:%.*]] = load i8*, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 474 // CHECK1-NEXT: [[TMP15:%.*]] = bitcast void (i8*, ...)* [[TMP13]] to void (i8*, i8***)* 475 // CHECK1-NEXT: call void [[TMP15]](i8* [[TMP14]], i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR5:[0-9]+]] 476 // CHECK1-NEXT: [[TMP16:%.*]] = load i8**, i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias !12 477 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_ANON_0:%.*]], %struct.anon.0* [[TMP12]], i32 0, i32 1 478 // CHECK1-NEXT: [[TMP18:%.*]] = load i32*, i32** [[TMP17]], align 8 479 // CHECK1-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP16]], align 8 480 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 481 // CHECK1-NEXT: [[TMP21:%.*]] = bitcast i32* [[TMP18]] to i8* 482 // CHECK1-NEXT: [[TMP22:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP19]], i8* [[TMP21]]) #[[ATTR5]] 483 // CHECK1-NEXT: [[CONV_I:%.*]] = bitcast i8* [[TMP22]] to i32* 484 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[TMP12]], i32 0, i32 2 485 // CHECK1-NEXT: [[TMP24:%.*]] = load i8**, i8*** [[TMP23]], align 8 486 // CHECK1-NEXT: [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8 487 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[TMP12]], i32 0, i32 1 488 // CHECK1-NEXT: [[TMP27:%.*]] = load i32*, i32** [[TMP26]], align 8 489 // CHECK1-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4 490 // CHECK1-NEXT: [[TMP29:%.*]] = sext i32 [[TMP28]] to i64 491 // CHECK1-NEXT: [[LB_ADD_LEN_I:%.*]] = add nsw i64 -1, [[TMP29]] 492 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[TMP12]], i32 0, i32 2 493 // CHECK1-NEXT: [[TMP31:%.*]] = load i8**, i8*** [[TMP30]], align 8 494 // CHECK1-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds i8*, i8** [[TMP31]], i64 9 495 // CHECK1-NEXT: [[TMP32:%.*]] = load i8*, i8** [[ARRAYIDX2_I]], align 8 496 // CHECK1-NEXT: [[ARRAYIDX3_I:%.*]] = getelementptr inbounds i8, i8* [[TMP32]], i64 [[LB_ADD_LEN_I]] 497 // CHECK1-NEXT: [[TMP33:%.*]] = ptrtoint i8* [[ARRAYIDX3_I]] to i64 498 // CHECK1-NEXT: [[TMP34:%.*]] = ptrtoint i8* [[TMP25]] to i64 499 // CHECK1-NEXT: [[TMP35:%.*]] = sub i64 [[TMP33]], [[TMP34]] 500 // CHECK1-NEXT: [[TMP36:%.*]] = sdiv exact i64 [[TMP35]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 501 // CHECK1-NEXT: [[TMP37:%.*]] = add nuw i64 [[TMP36]], 1 502 // CHECK1-NEXT: [[TMP38:%.*]] = mul nuw i64 [[TMP37]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 503 // CHECK1-NEXT: store i64 [[TMP37]], i64* @{{reduction_size[.].+[.]}}, align 8 504 // CHECK1-NEXT: [[TMP39:%.*]] = load i8*, i8** [[TMP16]], align 8 505 // CHECK1-NEXT: [[TMP40:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP39]], i8* [[TMP25]]) #[[ATTR5]] 506 // CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[TMP12]], i32 0, i32 2 507 // CHECK1-NEXT: [[TMP42:%.*]] = load i8**, i8*** [[TMP41]], align 8 508 // CHECK1-NEXT: [[TMP43:%.*]] = load i8*, i8** [[TMP42]], align 8 509 // CHECK1-NEXT: [[TMP44:%.*]] = ptrtoint i8* [[TMP43]] to i64 510 // CHECK1-NEXT: [[TMP45:%.*]] = ptrtoint i8* [[TMP25]] to i64 511 // CHECK1-NEXT: [[TMP46:%.*]] = sub i64 [[TMP44]], [[TMP45]] 512 // CHECK1-NEXT: [[TMP47:%.*]] = sdiv exact i64 [[TMP46]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 513 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr i8, i8* [[TMP40]], i64 [[TMP47]] 514 // CHECK1-NEXT: store i8** [[TMP4_I]], i8*** [[TMP_I]], align 8, !noalias !12 515 // CHECK1-NEXT: store i8* [[TMP48]], i8** [[TMP4_I]], align 8, !noalias !12 516 // CHECK1-NEXT: ret i32 0 517 // 518 // 519 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func 520 // CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 521 // CHECK1-NEXT: entry: 522 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 523 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 524 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 525 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 526 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 527 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [3 x i8*]* 528 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 529 // CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [3 x i8*]* 530 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 0 531 // CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 532 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 533 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 0 534 // CHECK1-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 535 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 536 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 1 537 // CHECK1-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 538 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1 539 // CHECK1-NEXT: [[TMP15:%.*]] = load i8*, i8** [[TMP14]], align 8 540 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2 541 // CHECK1-NEXT: [[TMP17:%.*]] = load i8*, i8** [[TMP16]], align 8 542 // CHECK1-NEXT: [[TMP18:%.*]] = ptrtoint i8* [[TMP17]] to i64 543 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP11]], align 4 544 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP8]], align 4 545 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 546 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4 547 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, i8* [[TMP15]], i64 [[TMP18]] 548 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP15]], [[TMP21]] 549 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 550 // CHECK1: omp.arraycpy.body: 551 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP13]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 552 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP15]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 553 // CHECK1-NEXT: [[TMP22:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 554 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP22]] to i32 555 // CHECK1-NEXT: [[TMP23:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 556 // CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP23]] to i32 557 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 558 // CHECK1-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8 559 // CHECK1-NEXT: store i8 [[CONV4]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 560 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 561 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 562 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP21]] 563 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]] 564 // CHECK1: omp.arraycpy.done5: 565 // CHECK1-NEXT: ret void 566 // 567 // 568 // CHECK2-LABEL: define {{[^@]+}}@main 569 // CHECK2-SAME: (i32 [[ARGC:%.*]], i8** [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 570 // CHECK2-NEXT: entry: 571 // CHECK2-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 572 // CHECK2-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 573 // CHECK2-NEXT: [[OMP_OUTLINED_ARG_AGG_:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8 574 // CHECK2-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 575 // CHECK2-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 576 // CHECK2-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[OMP_OUTLINED_ARG_AGG_]], i32 0, i32 0 577 // CHECK2-NEXT: store i32* [[ARGC_ADDR]], i32** [[TMP0]], align 8 578 // CHECK2-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[OMP_OUTLINED_ARG_AGG_]], i32 0, i32 1 579 // CHECK2-NEXT: [[TMP2:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 580 // CHECK2-NEXT: store i8** [[TMP2]], i8*** [[TMP1]], align 8 581 // CHECK2-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.anon*)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.anon* [[OMP_OUTLINED_ARG_AGG_]]) 582 // CHECK2-NEXT: ret i32 0 583 // 584 // 585 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined. 586 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.anon* noalias [[__CONTEXT:%.*]]) #[[ATTR1:[0-9]+]] { 587 // CHECK2-NEXT: entry: 588 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 589 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 590 // CHECK2-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca %struct.anon*, align 8 591 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 592 // CHECK2-NEXT: [[TMP:%.*]] = alloca i64, align 8 593 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8 594 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8 595 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8 596 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 597 // CHECK2-NEXT: [[ARGC:%.*]] = alloca i32, align 4 598 // CHECK2-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 599 // CHECK2-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 600 // CHECK2-NEXT: [[_TMP4:%.*]] = alloca i8**, align 8 601 // CHECK2-NEXT: [[_TMP5:%.*]] = alloca i8*, align 8 602 // CHECK2-NEXT: [[DOTRD_INPUT_:%.*]] = alloca [2 x %struct.kmp_taskred_input_t], align 8 603 // CHECK2-NEXT: [[DOTTASK_RED_:%.*]] = alloca i8*, align 8 604 // CHECK2-NEXT: [[I:%.*]] = alloca i64, align 8 605 // CHECK2-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_0:%.*]], align 8 606 // CHECK2-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8 607 // CHECK2-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1 608 // CHECK2-NEXT: [[_TMP27:%.*]] = alloca i8, align 1 609 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 610 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 611 // CHECK2-NEXT: store %struct.anon* [[__CONTEXT]], %struct.anon** [[__CONTEXT_ADDR]], align 8 612 // CHECK2-NEXT: [[TMP0:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR]], align 8 613 // CHECK2-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP0]], i32 0, i32 0 614 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[TMP1]], align 8 615 // CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP0]], i32 0, i32 1 616 // CHECK2-NEXT: store i64 0, i64* [[DOTOMP_LB]], align 8 617 // CHECK2-NEXT: store i64 9, i64* [[DOTOMP_UB]], align 8 618 // CHECK2-NEXT: store i64 1, i64* [[DOTOMP_STRIDE]], align 8 619 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 620 // CHECK2-NEXT: store i32 0, i32* [[ARGC]], align 4 621 // CHECK2-NEXT: [[TMP4:%.*]] = load i8**, i8*** [[TMP3]], align 8 622 // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8*, i8** [[TMP4]], i64 0 623 // CHECK2-NEXT: [[TMP5:%.*]] = load i8*, i8** [[ARRAYIDX]], align 8 624 // CHECK2-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, i8* [[TMP5]], i64 0 625 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP2]], align 4 626 // CHECK2-NEXT: [[TMP7:%.*]] = sext i32 [[TMP6]] to i64 627 // CHECK2-NEXT: [[LB_ADD_LEN:%.*]] = add nsw i64 -1, [[TMP7]] 628 // CHECK2-NEXT: [[TMP8:%.*]] = load i8**, i8*** [[TMP3]], align 8 629 // CHECK2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8*, i8** [[TMP8]], i64 9 630 // CHECK2-NEXT: [[TMP9:%.*]] = load i8*, i8** [[ARRAYIDX2]], align 8 631 // CHECK2-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds i8, i8* [[TMP9]], i64 [[LB_ADD_LEN]] 632 // CHECK2-NEXT: [[TMP10:%.*]] = ptrtoint i8* [[ARRAYIDX3]] to i64 633 // CHECK2-NEXT: [[TMP11:%.*]] = ptrtoint i8* [[ARRAYIDX1]] to i64 634 // CHECK2-NEXT: [[TMP12:%.*]] = sub i64 [[TMP10]], [[TMP11]] 635 // CHECK2-NEXT: [[TMP13:%.*]] = sdiv exact i64 [[TMP12]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 636 // CHECK2-NEXT: [[TMP14:%.*]] = add nuw i64 [[TMP13]], 1 637 // CHECK2-NEXT: [[TMP15:%.*]] = mul nuw i64 [[TMP14]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 638 // CHECK2-NEXT: [[TMP16:%.*]] = call i8* @llvm.stacksave() 639 // CHECK2-NEXT: store i8* [[TMP16]], i8** [[SAVED_STACK]], align 8 640 // CHECK2-NEXT: [[VLA:%.*]] = alloca i8, i64 [[TMP14]], align 16 641 // CHECK2-NEXT: store i64 [[TMP14]], i64* [[__VLA_EXPR0]], align 8 642 // CHECK2-NEXT: [[TMP17:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP14]] 643 // CHECK2-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[VLA]], [[TMP17]] 644 // CHECK2-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 645 // CHECK2: omp.arrayinit.body: 646 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 647 // CHECK2-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 648 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 649 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP17]] 650 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 651 // CHECK2: omp.arrayinit.done: 652 // CHECK2-NEXT: [[TMP18:%.*]] = load i8**, i8*** [[TMP3]], align 8 653 // CHECK2-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP18]], align 8 654 // CHECK2-NEXT: [[TMP20:%.*]] = ptrtoint i8* [[TMP19]] to i64 655 // CHECK2-NEXT: [[TMP21:%.*]] = ptrtoint i8* [[ARRAYIDX1]] to i64 656 // CHECK2-NEXT: [[TMP22:%.*]] = sub i64 [[TMP20]], [[TMP21]] 657 // CHECK2-NEXT: [[TMP23:%.*]] = sdiv exact i64 [[TMP22]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 658 // CHECK2-NEXT: [[TMP24:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP23]] 659 // CHECK2-NEXT: store i8** [[_TMP5]], i8*** [[_TMP4]], align 8 660 // CHECK2-NEXT: store i8* [[TMP24]], i8** [[_TMP5]], align 8 661 // CHECK2-NEXT: [[DOTRD_INPUT_GEP_:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]], i64 0, i64 0 662 // CHECK2-NEXT: [[TMP25:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T:%.*]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 0 663 // CHECK2-NEXT: [[TMP26:%.*]] = bitcast i32* [[ARGC]] to i8* 664 // CHECK2-NEXT: store i8* [[TMP26]], i8** [[TMP25]], align 8 665 // CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 1 666 // CHECK2-NEXT: [[TMP28:%.*]] = bitcast i32* [[TMP2]] to i8* 667 // CHECK2-NEXT: store i8* [[TMP28]], i8** [[TMP27]], align 8 668 // CHECK2-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 2 669 // CHECK2-NEXT: store i64 4, i64* [[TMP29]], align 8 670 // CHECK2-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 3 671 // CHECK2-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init. to i8*), i8** [[TMP30]], align 8 672 // CHECK2-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 4 673 // CHECK2-NEXT: store i8* null, i8** [[TMP31]], align 8 674 // CHECK2-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 5 675 // CHECK2-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb. to i8*), i8** [[TMP32]], align 8 676 // CHECK2-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 6 677 // CHECK2-NEXT: [[TMP34:%.*]] = bitcast i32* [[TMP33]] to i8* 678 // CHECK2-NEXT: call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP34]], i8 0, i64 4, i1 false) 679 // CHECK2-NEXT: [[DOTRD_INPUT_GEP_6:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]], i64 0, i64 1 680 // CHECK2-NEXT: [[TMP35:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 0 681 // CHECK2-NEXT: [[TMP36:%.*]] = load i8**, i8*** [[TMP3]], align 8 682 // CHECK2-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds i8*, i8** [[TMP36]], i64 0 683 // CHECK2-NEXT: [[TMP37:%.*]] = load i8*, i8** [[ARRAYIDX7]], align 8 684 // CHECK2-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i8, i8* [[TMP37]], i64 0 685 // CHECK2-NEXT: [[TMP38:%.*]] = load i32, i32* [[TMP2]], align 4 686 // CHECK2-NEXT: [[TMP39:%.*]] = sext i32 [[TMP38]] to i64 687 // CHECK2-NEXT: [[LB_ADD_LEN9:%.*]] = add nsw i64 -1, [[TMP39]] 688 // CHECK2-NEXT: [[TMP40:%.*]] = load i8**, i8*** [[TMP3]], align 8 689 // CHECK2-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i8*, i8** [[TMP40]], i64 9 690 // CHECK2-NEXT: [[TMP41:%.*]] = load i8*, i8** [[ARRAYIDX10]], align 8 691 // CHECK2-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds i8, i8* [[TMP41]], i64 [[LB_ADD_LEN9]] 692 // CHECK2-NEXT: store i8* [[VLA]], i8** [[TMP35]], align 8 693 // CHECK2-NEXT: [[TMP42:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 1 694 // CHECK2-NEXT: store i8* [[ARRAYIDX8]], i8** [[TMP42]], align 8 695 // CHECK2-NEXT: [[TMP43:%.*]] = ptrtoint i8* [[ARRAYIDX11]] to i64 696 // CHECK2-NEXT: [[TMP44:%.*]] = ptrtoint i8* [[ARRAYIDX8]] to i64 697 // CHECK2-NEXT: [[TMP45:%.*]] = sub i64 [[TMP43]], [[TMP44]] 698 // CHECK2-NEXT: [[TMP46:%.*]] = sdiv exact i64 [[TMP45]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 699 // CHECK2-NEXT: [[TMP47:%.*]] = add nuw i64 [[TMP46]], 1 700 // CHECK2-NEXT: [[TMP48:%.*]] = mul nuw i64 [[TMP47]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 701 // CHECK2-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 2 702 // CHECK2-NEXT: store i64 [[TMP48]], i64* [[TMP49]], align 8 703 // CHECK2-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 3 704 // CHECK2-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init..1 to i8*), i8** [[TMP50]], align 8 705 // CHECK2-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 4 706 // CHECK2-NEXT: store i8* null, i8** [[TMP51]], align 8 707 // CHECK2-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 5 708 // CHECK2-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb..2 to i8*), i8** [[TMP52]], align 8 709 // CHECK2-NEXT: [[TMP53:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 6 710 // CHECK2-NEXT: store i32 1, i32* [[TMP53]], align 8 711 // CHECK2-NEXT: [[TMP54:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 712 // CHECK2-NEXT: [[TMP55:%.*]] = load i32, i32* [[TMP54]], align 4 713 // CHECK2-NEXT: [[TMP56:%.*]] = bitcast [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]] to i8* 714 // CHECK2-NEXT: [[TMP57:%.*]] = call i8* @__kmpc_taskred_modifier_init(%struct.ident_t* @[[GLOB1]], i32 [[TMP55]], i32 1, i32 2, i8* [[TMP56]]) 715 // CHECK2-NEXT: store i8* [[TMP57]], i8** [[DOTTASK_RED_]], align 8 716 // CHECK2-NEXT: [[TMP58:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 717 // CHECK2-NEXT: [[TMP59:%.*]] = load i32, i32* [[TMP58]], align 4 718 // CHECK2-NEXT: call void @__kmpc_for_static_init_8(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP59]], i32 34, i32* [[DOTOMP_IS_LAST]], i64* [[DOTOMP_LB]], i64* [[DOTOMP_UB]], i64* [[DOTOMP_STRIDE]], i64 1, i64 1) 719 // CHECK2-NEXT: [[TMP60:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 720 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i64 [[TMP60]], 9 721 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 722 // CHECK2: cond.true: 723 // CHECK2-NEXT: br label [[COND_END:%.*]] 724 // CHECK2: cond.false: 725 // CHECK2-NEXT: [[TMP61:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 726 // CHECK2-NEXT: br label [[COND_END]] 727 // CHECK2: cond.end: 728 // CHECK2-NEXT: [[COND:%.*]] = phi i64 [ 9, [[COND_TRUE]] ], [ [[TMP61]], [[COND_FALSE]] ] 729 // CHECK2-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8 730 // CHECK2-NEXT: [[TMP62:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8 731 // CHECK2-NEXT: store i64 [[TMP62]], i64* [[DOTOMP_IV]], align 8 732 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 733 // CHECK2: omp.inner.for.cond: 734 // CHECK2-NEXT: [[TMP63:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 735 // CHECK2-NEXT: [[TMP64:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 736 // CHECK2-NEXT: [[CMP12:%.*]] = icmp sle i64 [[TMP63]], [[TMP64]] 737 // CHECK2-NEXT: br i1 [[CMP12]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]] 738 // CHECK2: omp.inner.for.cond.cleanup: 739 // CHECK2-NEXT: br label [[OMP_INNER_FOR_END:%.*]] 740 // CHECK2: omp.inner.for.body: 741 // CHECK2-NEXT: [[TMP65:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 742 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i64 [[TMP65]], 1 743 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL]] 744 // CHECK2-NEXT: store i64 [[ADD]], i64* [[I]], align 8 745 // CHECK2-NEXT: [[TMP66:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[AGG_CAPTURED]], i32 0, i32 0 746 // CHECK2-NEXT: store i8** [[DOTTASK_RED_]], i8*** [[TMP66]], align 8 747 // CHECK2-NEXT: [[TMP67:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[AGG_CAPTURED]], i32 0, i32 1 748 // CHECK2-NEXT: store i32* [[ARGC]], i32** [[TMP67]], align 8 749 // CHECK2-NEXT: [[TMP68:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[AGG_CAPTURED]], i32 0, i32 2 750 // CHECK2-NEXT: [[TMP69:%.*]] = load i8**, i8*** [[_TMP4]], align 8 751 // CHECK2-NEXT: store i8** [[TMP69]], i8*** [[TMP68]], align 8 752 // CHECK2-NEXT: [[TMP70:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 753 // CHECK2-NEXT: [[TMP71:%.*]] = load i32, i32* [[TMP70]], align 4 754 // CHECK2-NEXT: [[TMP72:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP71]], i32 1, i64 48, i64 24, i32 (i32, i8*)* bitcast (i32 (i32, %struct.kmp_task_t_with_privates*)* @.omp_task_entry. to i32 (i32, i8*)*)) 755 // CHECK2-NEXT: [[TMP73:%.*]] = bitcast i8* [[TMP72]] to %struct.kmp_task_t_with_privates* 756 // CHECK2-NEXT: [[TMP74:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP73]], i32 0, i32 0 757 // CHECK2-NEXT: [[TMP75:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP74]], i32 0, i32 0 758 // CHECK2-NEXT: [[TMP76:%.*]] = load i8*, i8** [[TMP75]], align 8 759 // CHECK2-NEXT: [[TMP77:%.*]] = bitcast %struct.anon.0* [[AGG_CAPTURED]] to i8* 760 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP76]], i8* align 8 [[TMP77]], i64 24, i1 false) 761 // CHECK2-NEXT: [[TMP78:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP73]], i32 0, i32 1 762 // CHECK2-NEXT: [[TMP79:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP78]], i32 0, i32 0 763 // CHECK2-NEXT: [[TMP80:%.*]] = load i8*, i8** [[DOTTASK_RED_]], align 8 764 // CHECK2-NEXT: store i8* [[TMP80]], i8** [[TMP79]], align 8 765 // CHECK2-NEXT: [[TMP81:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 766 // CHECK2-NEXT: [[TMP82:%.*]] = load i32, i32* [[TMP81]], align 4 767 // CHECK2-NEXT: [[TMP83:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP82]], i8* [[TMP72]]) 768 // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 769 // CHECK2: omp.body.continue: 770 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 771 // CHECK2: omp.inner.for.inc: 772 // CHECK2-NEXT: [[TMP84:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 773 // CHECK2-NEXT: [[ADD13:%.*]] = add nsw i64 [[TMP84]], 1 774 // CHECK2-NEXT: store i64 [[ADD13]], i64* [[DOTOMP_IV]], align 8 775 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] 776 // CHECK2: omp.inner.for.end: 777 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 778 // CHECK2: omp.loop.exit: 779 // CHECK2-NEXT: [[TMP85:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 780 // CHECK2-NEXT: [[TMP86:%.*]] = load i32, i32* [[TMP85]], align 4 781 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP86]]) 782 // CHECK2-NEXT: [[TMP87:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 783 // CHECK2-NEXT: [[TMP88:%.*]] = load i32, i32* [[TMP87]], align 4 784 // CHECK2-NEXT: call void @__kmpc_task_reduction_modifier_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP88]], i32 1) 785 // CHECK2-NEXT: [[TMP89:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 786 // CHECK2-NEXT: [[TMP90:%.*]] = bitcast i32* [[ARGC]] to i8* 787 // CHECK2-NEXT: store i8* [[TMP90]], i8** [[TMP89]], align 8 788 // CHECK2-NEXT: [[TMP91:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 789 // CHECK2-NEXT: store i8* [[VLA]], i8** [[TMP91]], align 8 790 // CHECK2-NEXT: [[TMP92:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 791 // CHECK2-NEXT: [[TMP93:%.*]] = inttoptr i64 [[TMP14]] to i8* 792 // CHECK2-NEXT: store i8* [[TMP93]], i8** [[TMP92]], align 8 793 // CHECK2-NEXT: [[TMP94:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 794 // CHECK2-NEXT: [[TMP95:%.*]] = load i32, i32* [[TMP94]], align 4 795 // CHECK2-NEXT: [[TMP96:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 796 // CHECK2-NEXT: [[TMP97:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[TMP95]], i32 2, i64 24, i8* [[TMP96]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var) 797 // CHECK2-NEXT: switch i32 [[TMP97]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 798 // CHECK2-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 799 // CHECK2-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 800 // CHECK2-NEXT: ] 801 // CHECK2: .omp.reduction.case1: 802 // CHECK2-NEXT: [[TMP98:%.*]] = load i32, i32* [[TMP2]], align 4 803 // CHECK2-NEXT: [[TMP99:%.*]] = load i32, i32* [[ARGC]], align 4 804 // CHECK2-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP98]], [[TMP99]] 805 // CHECK2-NEXT: store i32 [[ADD14]], i32* [[TMP2]], align 4 806 // CHECK2-NEXT: [[TMP100:%.*]] = getelementptr i8, i8* [[ARRAYIDX1]], i64 [[TMP14]] 807 // CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[ARRAYIDX1]], [[TMP100]] 808 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE21:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 809 // CHECK2: omp.arraycpy.body: 810 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 811 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST15:%.*]] = phi i8* [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT19:%.*]], [[OMP_ARRAYCPY_BODY]] ] 812 // CHECK2-NEXT: [[TMP101:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST15]], align 1 813 // CHECK2-NEXT: [[CONV:%.*]] = sext i8 [[TMP101]] to i32 814 // CHECK2-NEXT: [[TMP102:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 815 // CHECK2-NEXT: [[CONV16:%.*]] = sext i8 [[TMP102]] to i32 816 // CHECK2-NEXT: [[ADD17:%.*]] = add nsw i32 [[CONV]], [[CONV16]] 817 // CHECK2-NEXT: [[CONV18:%.*]] = trunc i32 [[ADD17]] to i8 818 // CHECK2-NEXT: store i8 [[CONV18]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST15]], align 1 819 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT19]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST15]], i32 1 820 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 821 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE20:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT19]], [[TMP100]] 822 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE20]], label [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_BODY]] 823 // CHECK2: omp.arraycpy.done21: 824 // CHECK2-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP95]], [8 x i32]* @.gomp_critical_user_.reduction.var) 825 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 826 // CHECK2: .omp.reduction.case2: 827 // CHECK2-NEXT: [[TMP103:%.*]] = load i32, i32* [[ARGC]], align 4 828 // CHECK2-NEXT: [[TMP104:%.*]] = atomicrmw add i32* [[TMP2]], i32 [[TMP103]] monotonic, align 4 829 // CHECK2-NEXT: [[TMP105:%.*]] = getelementptr i8, i8* [[ARRAYIDX1]], i64 [[TMP14]] 830 // CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY22:%.*]] = icmp eq i8* [[ARRAYIDX1]], [[TMP105]] 831 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY22]], label [[OMP_ARRAYCPY_DONE35:%.*]], label [[OMP_ARRAYCPY_BODY23:%.*]] 832 // CHECK2: omp.arraycpy.body23: 833 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST24:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT33:%.*]], [[ATOMIC_EXIT:%.*]] ] 834 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST25:%.*]] = phi i8* [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT32:%.*]], [[ATOMIC_EXIT]] ] 835 // CHECK2-NEXT: [[TMP106:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST24]], align 1 836 // CHECK2-NEXT: [[CONV26:%.*]] = sext i8 [[TMP106]] to i32 837 // CHECK2-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST25]] monotonic, align 1 838 // CHECK2-NEXT: br label [[ATOMIC_CONT:%.*]] 839 // CHECK2: atomic_cont: 840 // CHECK2-NEXT: [[TMP107:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY23]] ], [ [[TMP112:%.*]], [[ATOMIC_CONT]] ] 841 // CHECK2-NEXT: store i8 [[TMP107]], i8* [[_TMP27]], align 1 842 // CHECK2-NEXT: [[TMP108:%.*]] = load i8, i8* [[_TMP27]], align 1 843 // CHECK2-NEXT: [[CONV28:%.*]] = sext i8 [[TMP108]] to i32 844 // CHECK2-NEXT: [[TMP109:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST24]], align 1 845 // CHECK2-NEXT: [[CONV29:%.*]] = sext i8 [[TMP109]] to i32 846 // CHECK2-NEXT: [[ADD30:%.*]] = add nsw i32 [[CONV28]], [[CONV29]] 847 // CHECK2-NEXT: [[CONV31:%.*]] = trunc i32 [[ADD30]] to i8 848 // CHECK2-NEXT: store i8 [[CONV31]], i8* [[ATOMIC_TEMP]], align 1 849 // CHECK2-NEXT: [[TMP110:%.*]] = load i8, i8* [[ATOMIC_TEMP]], align 1 850 // CHECK2-NEXT: [[TMP111:%.*]] = cmpxchg i8* [[OMP_ARRAYCPY_DESTELEMENTPAST25]], i8 [[TMP107]], i8 [[TMP110]] monotonic monotonic, align 1 851 // CHECK2-NEXT: [[TMP112]] = extractvalue { i8, i1 } [[TMP111]], 0 852 // CHECK2-NEXT: [[TMP113:%.*]] = extractvalue { i8, i1 } [[TMP111]], 1 853 // CHECK2-NEXT: br i1 [[TMP113]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]] 854 // CHECK2: atomic_exit: 855 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT32]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST25]], i32 1 856 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT33]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST24]], i32 1 857 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE34:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT32]], [[TMP105]] 858 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE34]], label [[OMP_ARRAYCPY_DONE35]], label [[OMP_ARRAYCPY_BODY23]] 859 // CHECK2: omp.arraycpy.done35: 860 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 861 // CHECK2: .omp.reduction.default: 862 // CHECK2-NEXT: [[TMP114:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 863 // CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP114]]) 864 // CHECK2-NEXT: ret void 865 // 866 // 867 // CHECK2-LABEL: define {{[^@]+}}@.red_init. 868 // CHECK2-SAME: (i8* noalias [[TMP0:%.*]], i8* noalias [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { 869 // CHECK2-NEXT: entry: 870 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 871 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 872 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 873 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 874 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 875 // CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to i32* 876 // CHECK2-NEXT: store i32 0, i32* [[TMP3]], align 8 877 // CHECK2-NEXT: ret void 878 // 879 // 880 // CHECK2-LABEL: define {{[^@]+}}@.red_comb. 881 // CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 882 // CHECK2-NEXT: entry: 883 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 884 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 885 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 886 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 887 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 888 // CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to i32* 889 // CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 890 // CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to i32* 891 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP3]], align 8 892 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP5]], align 8 893 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP6]], [[TMP7]] 894 // CHECK2-NEXT: store i32 [[ADD]], i32* [[TMP3]], align 8 895 // CHECK2-NEXT: ret void 896 // 897 // 898 // CHECK2-LABEL: define {{[^@]+}}@.red_init..1 899 // CHECK2-SAME: (i8* noalias [[TMP0:%.*]], i8* noalias [[TMP1:%.*]]) #[[ATTR3]] { 900 // CHECK2-NEXT: entry: 901 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 902 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 903 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 904 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 905 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 906 // CHECK2-NEXT: [[TMP3:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8 907 // CHECK2-NEXT: [[TMP4:%.*]] = getelementptr i8, i8* [[TMP2]], i64 [[TMP3]] 908 // CHECK2-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[TMP2]], [[TMP4]] 909 // CHECK2-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 910 // CHECK2: omp.arrayinit.body: 911 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP2]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 912 // CHECK2-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 913 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 914 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]] 915 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 916 // CHECK2: omp.arrayinit.done: 917 // CHECK2-NEXT: ret void 918 // 919 // 920 // CHECK2-LABEL: define {{[^@]+}}@.red_comb..2 921 // CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 922 // CHECK2-NEXT: entry: 923 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 924 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 925 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 926 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 927 // CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8 928 // CHECK2-NEXT: [[TMP3:%.*]] = load i8*, i8** [[DOTADDR]], align 8 929 // CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 930 // CHECK2-NEXT: [[TMP5:%.*]] = getelementptr i8, i8* [[TMP3]], i64 [[TMP2]] 931 // CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP3]], [[TMP5]] 932 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 933 // CHECK2: omp.arraycpy.body: 934 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP4]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 935 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP3]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 936 // CHECK2-NEXT: [[TMP6:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 937 // CHECK2-NEXT: [[CONV:%.*]] = sext i8 [[TMP6]] to i32 938 // CHECK2-NEXT: [[TMP7:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 939 // CHECK2-NEXT: [[CONV2:%.*]] = sext i8 [[TMP7]] to i32 940 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 941 // CHECK2-NEXT: [[CONV3:%.*]] = trunc i32 [[ADD]] to i8 942 // CHECK2-NEXT: store i8 [[CONV3]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 943 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 944 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 945 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP5]] 946 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]] 947 // CHECK2: omp.arraycpy.done4: 948 // CHECK2-NEXT: ret void 949 // 950 // 951 // CHECK2-LABEL: define {{[^@]+}}@.omp_task_privates_map. 952 // CHECK2-SAME: (%struct..kmp_privates.t* noalias [[TMP0:%.*]], i8*** noalias [[TMP1:%.*]]) #[[ATTR6:[0-9]+]] { 953 // CHECK2-NEXT: entry: 954 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca %struct..kmp_privates.t*, align 8 955 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8***, align 8 956 // CHECK2-NEXT: store %struct..kmp_privates.t* [[TMP0]], %struct..kmp_privates.t** [[DOTADDR]], align 8 957 // CHECK2-NEXT: store i8*** [[TMP1]], i8**** [[DOTADDR1]], align 8 958 // CHECK2-NEXT: [[TMP2:%.*]] = load %struct..kmp_privates.t*, %struct..kmp_privates.t** [[DOTADDR]], align 8 959 // CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP2]], i32 0, i32 0 960 // CHECK2-NEXT: [[TMP4:%.*]] = load i8***, i8**** [[DOTADDR1]], align 8 961 // CHECK2-NEXT: store i8** [[TMP3]], i8*** [[TMP4]], align 8 962 // CHECK2-NEXT: ret void 963 // 964 // 965 // CHECK2-LABEL: define {{[^@]+}}@.omp_task_entry. 966 // CHECK2-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias [[TMP1:%.*]]) #[[ATTR3]] { 967 // CHECK2-NEXT: entry: 968 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4 969 // CHECK2-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8 970 // CHECK2-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8 971 // CHECK2-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8 972 // CHECK2-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8 973 // CHECK2-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon.0*, align 8 974 // CHECK2-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca i8**, align 8 975 // CHECK2-NEXT: [[TMP_I:%.*]] = alloca i8**, align 8 976 // CHECK2-NEXT: [[TMP4_I:%.*]] = alloca i8*, align 8 977 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4 978 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8 979 // CHECK2-NEXT: store i32 [[TMP0]], i32* [[DOTADDR]], align 4 980 // CHECK2-NEXT: store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8 981 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4 982 // CHECK2-NEXT: [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8 983 // CHECK2-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0 984 // CHECK2-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2 985 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0 986 // CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 987 // CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon.0* 988 // CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1 989 // CHECK2-NEXT: [[TMP10:%.*]] = bitcast %struct..kmp_privates.t* [[TMP9]] to i8* 990 // CHECK2-NEXT: [[TMP11:%.*]] = bitcast %struct.kmp_task_t_with_privates* [[TMP3]] to i8* 991 // CHECK2-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META3:![0-9]+]]) 992 // CHECK2-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META6:![0-9]+]]) 993 // CHECK2-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META8:![0-9]+]]) 994 // CHECK2-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]]) 995 // CHECK2-NEXT: store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 996 // CHECK2-NEXT: store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !12 997 // CHECK2-NEXT: store i8* [[TMP10]], i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 998 // CHECK2-NEXT: store void (i8*, ...)* bitcast (void (%struct..kmp_privates.t*, i8***)* @.omp_task_privates_map. to void (i8*, ...)*), void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12 999 // CHECK2-NEXT: store i8* [[TMP11]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !12 1000 // CHECK2-NEXT: store %struct.anon.0* [[TMP8]], %struct.anon.0** [[__CONTEXT_ADDR_I]], align 8, !noalias !12 1001 // CHECK2-NEXT: [[TMP12:%.*]] = load %struct.anon.0*, %struct.anon.0** [[__CONTEXT_ADDR_I]], align 8, !noalias !12 1002 // CHECK2-NEXT: [[TMP13:%.*]] = load void (i8*, ...)*, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12 1003 // CHECK2-NEXT: [[TMP14:%.*]] = load i8*, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 1004 // CHECK2-NEXT: [[TMP15:%.*]] = bitcast void (i8*, ...)* [[TMP13]] to void (i8*, i8***)* 1005 // CHECK2-NEXT: call void [[TMP15]](i8* [[TMP14]], i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR5:[0-9]+]] 1006 // CHECK2-NEXT: [[TMP16:%.*]] = load i8**, i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias !12 1007 // CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_ANON_0:%.*]], %struct.anon.0* [[TMP12]], i32 0, i32 1 1008 // CHECK2-NEXT: [[TMP18:%.*]] = load i32*, i32** [[TMP17]], align 8 1009 // CHECK2-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP16]], align 8 1010 // CHECK2-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 1011 // CHECK2-NEXT: [[TMP21:%.*]] = bitcast i32* [[TMP18]] to i8* 1012 // CHECK2-NEXT: [[TMP22:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP19]], i8* [[TMP21]]) #[[ATTR5]] 1013 // CHECK2-NEXT: [[CONV_I:%.*]] = bitcast i8* [[TMP22]] to i32* 1014 // CHECK2-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[TMP12]], i32 0, i32 2 1015 // CHECK2-NEXT: [[TMP24:%.*]] = load i8**, i8*** [[TMP23]], align 8 1016 // CHECK2-NEXT: [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8 1017 // CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[TMP12]], i32 0, i32 1 1018 // CHECK2-NEXT: [[TMP27:%.*]] = load i32*, i32** [[TMP26]], align 8 1019 // CHECK2-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4 1020 // CHECK2-NEXT: [[TMP29:%.*]] = sext i32 [[TMP28]] to i64 1021 // CHECK2-NEXT: [[LB_ADD_LEN_I:%.*]] = add nsw i64 -1, [[TMP29]] 1022 // CHECK2-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[TMP12]], i32 0, i32 2 1023 // CHECK2-NEXT: [[TMP31:%.*]] = load i8**, i8*** [[TMP30]], align 8 1024 // CHECK2-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds i8*, i8** [[TMP31]], i64 9 1025 // CHECK2-NEXT: [[TMP32:%.*]] = load i8*, i8** [[ARRAYIDX2_I]], align 8 1026 // CHECK2-NEXT: [[ARRAYIDX3_I:%.*]] = getelementptr inbounds i8, i8* [[TMP32]], i64 [[LB_ADD_LEN_I]] 1027 // CHECK2-NEXT: [[TMP33:%.*]] = ptrtoint i8* [[ARRAYIDX3_I]] to i64 1028 // CHECK2-NEXT: [[TMP34:%.*]] = ptrtoint i8* [[TMP25]] to i64 1029 // CHECK2-NEXT: [[TMP35:%.*]] = sub i64 [[TMP33]], [[TMP34]] 1030 // CHECK2-NEXT: [[TMP36:%.*]] = sdiv exact i64 [[TMP35]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 1031 // CHECK2-NEXT: [[TMP37:%.*]] = add nuw i64 [[TMP36]], 1 1032 // CHECK2-NEXT: [[TMP38:%.*]] = mul nuw i64 [[TMP37]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 1033 // CHECK2-NEXT: store i64 [[TMP37]], i64* @{{reduction_size[.].+[.]}}, align 8 1034 // CHECK2-NEXT: [[TMP39:%.*]] = load i8*, i8** [[TMP16]], align 8 1035 // CHECK2-NEXT: [[TMP40:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP39]], i8* [[TMP25]]) #[[ATTR5]] 1036 // CHECK2-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[TMP12]], i32 0, i32 2 1037 // CHECK2-NEXT: [[TMP42:%.*]] = load i8**, i8*** [[TMP41]], align 8 1038 // CHECK2-NEXT: [[TMP43:%.*]] = load i8*, i8** [[TMP42]], align 8 1039 // CHECK2-NEXT: [[TMP44:%.*]] = ptrtoint i8* [[TMP43]] to i64 1040 // CHECK2-NEXT: [[TMP45:%.*]] = ptrtoint i8* [[TMP25]] to i64 1041 // CHECK2-NEXT: [[TMP46:%.*]] = sub i64 [[TMP44]], [[TMP45]] 1042 // CHECK2-NEXT: [[TMP47:%.*]] = sdiv exact i64 [[TMP46]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 1043 // CHECK2-NEXT: [[TMP48:%.*]] = getelementptr i8, i8* [[TMP40]], i64 [[TMP47]] 1044 // CHECK2-NEXT: store i8** [[TMP4_I]], i8*** [[TMP_I]], align 8, !noalias !12 1045 // CHECK2-NEXT: store i8* [[TMP48]], i8** [[TMP4_I]], align 8, !noalias !12 1046 // CHECK2-NEXT: ret i32 0 1047 // 1048 // 1049 // CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func 1050 // CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 1051 // CHECK2-NEXT: entry: 1052 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 1053 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 1054 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 1055 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 1056 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 1057 // CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [3 x i8*]* 1058 // CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 1059 // CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [3 x i8*]* 1060 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 0 1061 // CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 1062 // CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 1063 // CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 0 1064 // CHECK2-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 1065 // CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 1066 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 1 1067 // CHECK2-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 1068 // CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1 1069 // CHECK2-NEXT: [[TMP15:%.*]] = load i8*, i8** [[TMP14]], align 8 1070 // CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2 1071 // CHECK2-NEXT: [[TMP17:%.*]] = load i8*, i8** [[TMP16]], align 8 1072 // CHECK2-NEXT: [[TMP18:%.*]] = ptrtoint i8* [[TMP17]] to i64 1073 // CHECK2-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP11]], align 4 1074 // CHECK2-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP8]], align 4 1075 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 1076 // CHECK2-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4 1077 // CHECK2-NEXT: [[TMP21:%.*]] = getelementptr i8, i8* [[TMP15]], i64 [[TMP18]] 1078 // CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP15]], [[TMP21]] 1079 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 1080 // CHECK2: omp.arraycpy.body: 1081 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP13]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 1082 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP15]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 1083 // CHECK2-NEXT: [[TMP22:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 1084 // CHECK2-NEXT: [[CONV:%.*]] = sext i8 [[TMP22]] to i32 1085 // CHECK2-NEXT: [[TMP23:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 1086 // CHECK2-NEXT: [[CONV2:%.*]] = sext i8 [[TMP23]] to i32 1087 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 1088 // CHECK2-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8 1089 // CHECK2-NEXT: store i8 [[CONV4]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 1090 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 1091 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 1092 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP21]] 1093 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]] 1094 // CHECK2: omp.arraycpy.done5: 1095 // CHECK2-NEXT: ret void 1096 // 1097