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 -no-opaque-pointers -verify -fopenmp -fopenmp-version=50 -x c++ -triple x86_64-unknown-linux -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s --check-prefix=CHECK1 3 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=50 -x c++ -std=c++11 -triple x86_64-unknown-linux -fexceptions -fcxx-exceptions -emit-pch -o %t %s 4 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp -fopenmp-version=50 -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=CHECK1 5 6 // RUN: %clang_cc1 -no-opaque-pointers -triple x86_64-unknown-linux -verify -fopenmp-simd -fopenmp-version=50 -x c++ -emit-llvm %s -fexceptions -fcxx-exceptions -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 7 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -fopenmp-version=50 -x c++ -std=c++11 -triple x86_64-unknown-unknown -fexceptions -fcxx-exceptions -emit-pch -o %t %s 8 // RUN: %clang_cc1 -no-opaque-pointers -fopenmp-simd -fopenmp-version=50 -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 target parallel reduction(task, +: argc, argv[0:10][0:argc]) 15 { 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 noundef [[ARGC:%.*]], i8** noundef [[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: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 44 // CHECK1-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 45 // CHECK1-NEXT: [[TMP0:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 46 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l14(i32* [[ARGC_ADDR]], i8** [[TMP0]]) #[[ATTR5:[0-9]+]] 47 // CHECK1-NEXT: ret i32 0 48 // 49 // 50 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l14 51 // CHECK1-SAME: (i32* noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR1:[0-9]+]] { 52 // CHECK1-NEXT: entry: 53 // CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca i32*, align 8 54 // CHECK1-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 55 // CHECK1-NEXT: store i32* [[ARGC]], i32** [[ARGC_ADDR]], align 8 56 // CHECK1-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 57 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARGC_ADDR]], align 8 58 // CHECK1-NEXT: [[TMP1:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 59 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i8**)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[TMP0]], i8** [[TMP1]]) 60 // CHECK1-NEXT: ret void 61 // 62 // 63 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined. 64 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR1]] { 65 // CHECK1-NEXT: entry: 66 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 67 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 68 // CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca i32*, align 8 69 // CHECK1-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 70 // CHECK1-NEXT: [[ARGC1:%.*]] = alloca i32, align 4 71 // CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 72 // CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 73 // CHECK1-NEXT: [[TMP:%.*]] = alloca i8**, align 8 74 // CHECK1-NEXT: [[_TMP5:%.*]] = alloca i8*, align 8 75 // CHECK1-NEXT: [[DOTRD_INPUT_:%.*]] = alloca [2 x %struct.kmp_taskred_input_t], align 8 76 // CHECK1-NEXT: [[DOTTASK_RED_:%.*]] = alloca i8*, align 8 77 // CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8 78 // CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8 79 // CHECK1-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1 80 // CHECK1-NEXT: [[_TMP24:%.*]] = alloca i8, align 1 81 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 82 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 83 // CHECK1-NEXT: store i32* [[ARGC]], i32** [[ARGC_ADDR]], align 8 84 // CHECK1-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 85 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARGC_ADDR]], align 8 86 // CHECK1-NEXT: store i32 0, i32* [[ARGC1]], align 4 87 // CHECK1-NEXT: [[TMP1:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 88 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8*, i8** [[TMP1]], i64 0 89 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[ARRAYIDX]], align 8 90 // CHECK1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, i8* [[TMP2]], i64 0 91 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 92 // CHECK1-NEXT: [[TMP4:%.*]] = sext i32 [[TMP3]] to i64 93 // CHECK1-NEXT: [[LB_ADD_LEN:%.*]] = add nsw i64 -1, [[TMP4]] 94 // CHECK1-NEXT: [[TMP5:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 95 // CHECK1-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds i8*, i8** [[TMP5]], i64 9 96 // CHECK1-NEXT: [[TMP6:%.*]] = load i8*, i8** [[ARRAYIDX3]], align 8 97 // CHECK1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i8, i8* [[TMP6]], i64 [[LB_ADD_LEN]] 98 // CHECK1-NEXT: [[TMP7:%.*]] = ptrtoint i8* [[ARRAYIDX4]] to i64 99 // CHECK1-NEXT: [[TMP8:%.*]] = ptrtoint i8* [[ARRAYIDX2]] to i64 100 // CHECK1-NEXT: [[TMP9:%.*]] = sub i64 [[TMP7]], [[TMP8]] 101 // CHECK1-NEXT: [[TMP10:%.*]] = sdiv exact i64 [[TMP9]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 102 // CHECK1-NEXT: [[TMP11:%.*]] = add nuw i64 [[TMP10]], 1 103 // CHECK1-NEXT: [[TMP12:%.*]] = mul nuw i64 [[TMP11]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 104 // CHECK1-NEXT: [[TMP13:%.*]] = call i8* @llvm.stacksave() 105 // CHECK1-NEXT: store i8* [[TMP13]], i8** [[SAVED_STACK]], align 8 106 // CHECK1-NEXT: [[VLA:%.*]] = alloca i8, i64 [[TMP11]], align 16 107 // CHECK1-NEXT: store i64 [[TMP11]], i64* [[__VLA_EXPR0]], align 8 108 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP11]] 109 // CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[VLA]], [[TMP14]] 110 // CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 111 // CHECK1: omp.arrayinit.body: 112 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 113 // CHECK1-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 114 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 115 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP14]] 116 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 117 // CHECK1: omp.arrayinit.done: 118 // CHECK1-NEXT: [[TMP15:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 119 // CHECK1-NEXT: [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8 120 // CHECK1-NEXT: [[TMP17:%.*]] = ptrtoint i8* [[TMP16]] to i64 121 // CHECK1-NEXT: [[TMP18:%.*]] = ptrtoint i8* [[ARRAYIDX2]] to i64 122 // CHECK1-NEXT: [[TMP19:%.*]] = sub i64 [[TMP17]], [[TMP18]] 123 // CHECK1-NEXT: [[TMP20:%.*]] = sdiv exact i64 [[TMP19]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 124 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP20]] 125 // CHECK1-NEXT: store i8** [[_TMP5]], i8*** [[TMP]], align 8 126 // CHECK1-NEXT: store i8* [[TMP21]], i8** [[_TMP5]], align 8 127 // 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 0 128 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T:%.*]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 0 129 // CHECK1-NEXT: [[TMP23:%.*]] = bitcast i32* [[ARGC1]] to i8* 130 // CHECK1-NEXT: store i8* [[TMP23]], i8** [[TMP22]], align 8 131 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 1 132 // CHECK1-NEXT: [[TMP25:%.*]] = bitcast i32* [[TMP0]] to i8* 133 // CHECK1-NEXT: store i8* [[TMP25]], i8** [[TMP24]], align 8 134 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 2 135 // CHECK1-NEXT: store i64 4, i64* [[TMP26]], align 8 136 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 3 137 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init. to i8*), i8** [[TMP27]], align 8 138 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 4 139 // CHECK1-NEXT: store i8* null, i8** [[TMP28]], align 8 140 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 5 141 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb. to i8*), i8** [[TMP29]], align 8 142 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 6 143 // CHECK1-NEXT: [[TMP31:%.*]] = bitcast i32* [[TMP30]] to i8* 144 // CHECK1-NEXT: call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP31]], i8 0, i64 4, i1 false) 145 // 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 1 146 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 0 147 // CHECK1-NEXT: [[TMP33:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 148 // CHECK1-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds i8*, i8** [[TMP33]], i64 0 149 // CHECK1-NEXT: [[TMP34:%.*]] = load i8*, i8** [[ARRAYIDX7]], align 8 150 // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i8, i8* [[TMP34]], i64 0 151 // CHECK1-NEXT: [[TMP35:%.*]] = load i32, i32* [[TMP0]], align 4 152 // CHECK1-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 153 // CHECK1-NEXT: [[LB_ADD_LEN9:%.*]] = add nsw i64 -1, [[TMP36]] 154 // CHECK1-NEXT: [[TMP37:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 155 // CHECK1-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i8*, i8** [[TMP37]], i64 9 156 // CHECK1-NEXT: [[TMP38:%.*]] = load i8*, i8** [[ARRAYIDX10]], align 8 157 // CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds i8, i8* [[TMP38]], i64 [[LB_ADD_LEN9]] 158 // CHECK1-NEXT: store i8* [[VLA]], i8** [[TMP32]], align 8 159 // CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 1 160 // CHECK1-NEXT: store i8* [[ARRAYIDX8]], i8** [[TMP39]], align 8 161 // CHECK1-NEXT: [[TMP40:%.*]] = ptrtoint i8* [[ARRAYIDX11]] to i64 162 // CHECK1-NEXT: [[TMP41:%.*]] = ptrtoint i8* [[ARRAYIDX8]] to i64 163 // CHECK1-NEXT: [[TMP42:%.*]] = sub i64 [[TMP40]], [[TMP41]] 164 // CHECK1-NEXT: [[TMP43:%.*]] = sdiv exact i64 [[TMP42]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 165 // CHECK1-NEXT: [[TMP44:%.*]] = add nuw i64 [[TMP43]], 1 166 // CHECK1-NEXT: [[TMP45:%.*]] = mul nuw i64 [[TMP44]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 167 // CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 2 168 // CHECK1-NEXT: store i64 [[TMP45]], i64* [[TMP46]], align 8 169 // CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 3 170 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init..1 to i8*), i8** [[TMP47]], align 8 171 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 4 172 // CHECK1-NEXT: store i8* null, i8** [[TMP48]], align 8 173 // CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 5 174 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb..2 to i8*), i8** [[TMP49]], align 8 175 // CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_6]], i32 0, i32 6 176 // CHECK1-NEXT: store i32 1, i32* [[TMP50]], align 8 177 // CHECK1-NEXT: [[TMP51:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 178 // CHECK1-NEXT: [[TMP52:%.*]] = load i32, i32* [[TMP51]], align 4 179 // CHECK1-NEXT: [[TMP53:%.*]] = bitcast [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]] to i8* 180 // CHECK1-NEXT: [[TMP54:%.*]] = call i8* @__kmpc_taskred_modifier_init(%struct.ident_t* @[[GLOB1]], i32 [[TMP52]], i32 0, i32 2, i8* [[TMP53]]) 181 // CHECK1-NEXT: store i8* [[TMP54]], i8** [[DOTTASK_RED_]], align 8 182 // CHECK1-NEXT: [[TMP55:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 0 183 // CHECK1-NEXT: store i8** [[DOTTASK_RED_]], i8*** [[TMP55]], align 8 184 // CHECK1-NEXT: [[TMP56:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 1 185 // CHECK1-NEXT: store i32* [[ARGC1]], i32** [[TMP56]], align 8 186 // CHECK1-NEXT: [[TMP57:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 2 187 // CHECK1-NEXT: [[TMP58:%.*]] = load i8**, i8*** [[TMP]], align 8 188 // CHECK1-NEXT: store i8** [[TMP58]], i8*** [[TMP57]], align 8 189 // CHECK1-NEXT: [[TMP59:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 190 // CHECK1-NEXT: [[TMP60:%.*]] = load i32, i32* [[TMP59]], align 4 191 // CHECK1-NEXT: [[TMP61:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP60]], 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*)*)) 192 // CHECK1-NEXT: [[TMP62:%.*]] = bitcast i8* [[TMP61]] to %struct.kmp_task_t_with_privates* 193 // CHECK1-NEXT: [[TMP63:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP62]], i32 0, i32 0 194 // CHECK1-NEXT: [[TMP64:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP63]], i32 0, i32 0 195 // CHECK1-NEXT: [[TMP65:%.*]] = load i8*, i8** [[TMP64]], align 8 196 // CHECK1-NEXT: [[TMP66:%.*]] = bitcast %struct.anon* [[AGG_CAPTURED]] to i8* 197 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP65]], i8* align 8 [[TMP66]], i64 24, i1 false) 198 // CHECK1-NEXT: [[TMP67:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP62]], i32 0, i32 1 199 // CHECK1-NEXT: [[TMP68:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP67]], i32 0, i32 0 200 // CHECK1-NEXT: [[TMP69:%.*]] = load i8*, i8** [[DOTTASK_RED_]], align 8 201 // CHECK1-NEXT: store i8* [[TMP69]], i8** [[TMP68]], align 8 202 // CHECK1-NEXT: [[TMP70:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 203 // CHECK1-NEXT: [[TMP71:%.*]] = load i32, i32* [[TMP70]], align 4 204 // CHECK1-NEXT: [[TMP72:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP71]], i8* [[TMP61]]) 205 // CHECK1-NEXT: [[TMP73:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 206 // CHECK1-NEXT: [[TMP74:%.*]] = load i32, i32* [[TMP73]], align 4 207 // CHECK1-NEXT: call void @__kmpc_task_reduction_modifier_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP74]], i32 0) 208 // CHECK1-NEXT: [[TMP75:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 209 // CHECK1-NEXT: [[TMP76:%.*]] = bitcast i32* [[ARGC1]] to i8* 210 // CHECK1-NEXT: store i8* [[TMP76]], i8** [[TMP75]], align 8 211 // CHECK1-NEXT: [[TMP77:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 212 // CHECK1-NEXT: store i8* [[VLA]], i8** [[TMP77]], align 8 213 // CHECK1-NEXT: [[TMP78:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 214 // CHECK1-NEXT: [[TMP79:%.*]] = inttoptr i64 [[TMP11]] to i8* 215 // CHECK1-NEXT: store i8* [[TMP79]], i8** [[TMP78]], align 8 216 // CHECK1-NEXT: [[TMP80:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 217 // CHECK1-NEXT: [[TMP81:%.*]] = load i32, i32* [[TMP80]], align 4 218 // CHECK1-NEXT: [[TMP82:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 219 // CHECK1-NEXT: [[TMP83:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP81]], i32 2, i64 24, i8* [[TMP82]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var) 220 // CHECK1-NEXT: switch i32 [[TMP83]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 221 // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 222 // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 223 // CHECK1-NEXT: ] 224 // CHECK1: .omp.reduction.case1: 225 // CHECK1-NEXT: [[TMP84:%.*]] = load i32, i32* [[TMP0]], align 4 226 // CHECK1-NEXT: [[TMP85:%.*]] = load i32, i32* [[ARGC1]], align 4 227 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP84]], [[TMP85]] 228 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP0]], align 4 229 // CHECK1-NEXT: [[TMP86:%.*]] = getelementptr i8, i8* [[ARRAYIDX2]], i64 [[TMP11]] 230 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[ARRAYIDX2]], [[TMP86]] 231 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE18:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 232 // CHECK1: omp.arraycpy.body: 233 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 234 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST12:%.*]] = phi i8* [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT16:%.*]], [[OMP_ARRAYCPY_BODY]] ] 235 // CHECK1-NEXT: [[TMP87:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST12]], align 1 236 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP87]] to i32 237 // CHECK1-NEXT: [[TMP88:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 238 // CHECK1-NEXT: [[CONV13:%.*]] = sext i8 [[TMP88]] to i32 239 // CHECK1-NEXT: [[ADD14:%.*]] = add nsw i32 [[CONV]], [[CONV13]] 240 // CHECK1-NEXT: [[CONV15:%.*]] = trunc i32 [[ADD14]] to i8 241 // CHECK1-NEXT: store i8 [[CONV15]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST12]], align 1 242 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT16]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST12]], i32 1 243 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 244 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE17:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT16]], [[TMP86]] 245 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE17]], label [[OMP_ARRAYCPY_DONE18]], label [[OMP_ARRAYCPY_BODY]] 246 // CHECK1: omp.arraycpy.done18: 247 // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB2]], i32 [[TMP81]], [8 x i32]* @.gomp_critical_user_.reduction.var) 248 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 249 // CHECK1: .omp.reduction.case2: 250 // CHECK1-NEXT: [[TMP89:%.*]] = load i32, i32* [[ARGC1]], align 4 251 // CHECK1-NEXT: [[TMP90:%.*]] = atomicrmw add i32* [[TMP0]], i32 [[TMP89]] monotonic, align 4 252 // CHECK1-NEXT: [[TMP91:%.*]] = getelementptr i8, i8* [[ARRAYIDX2]], i64 [[TMP11]] 253 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY19:%.*]] = icmp eq i8* [[ARRAYIDX2]], [[TMP91]] 254 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY19]], label [[OMP_ARRAYCPY_DONE32:%.*]], label [[OMP_ARRAYCPY_BODY20:%.*]] 255 // CHECK1: omp.arraycpy.body20: 256 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST21:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT30:%.*]], [[ATOMIC_EXIT:%.*]] ] 257 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST22:%.*]] = phi i8* [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT29:%.*]], [[ATOMIC_EXIT]] ] 258 // CHECK1-NEXT: [[TMP92:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST21]], align 1 259 // CHECK1-NEXT: [[CONV23:%.*]] = sext i8 [[TMP92]] to i32 260 // CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST22]] monotonic, align 1 261 // CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]] 262 // CHECK1: atomic_cont: 263 // CHECK1-NEXT: [[TMP93:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY20]] ], [ [[TMP98:%.*]], [[ATOMIC_CONT]] ] 264 // CHECK1-NEXT: store i8 [[TMP93]], i8* [[_TMP24]], align 1 265 // CHECK1-NEXT: [[TMP94:%.*]] = load i8, i8* [[_TMP24]], align 1 266 // CHECK1-NEXT: [[CONV25:%.*]] = sext i8 [[TMP94]] to i32 267 // CHECK1-NEXT: [[TMP95:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST21]], align 1 268 // CHECK1-NEXT: [[CONV26:%.*]] = sext i8 [[TMP95]] to i32 269 // CHECK1-NEXT: [[ADD27:%.*]] = add nsw i32 [[CONV25]], [[CONV26]] 270 // CHECK1-NEXT: [[CONV28:%.*]] = trunc i32 [[ADD27]] to i8 271 // CHECK1-NEXT: store i8 [[CONV28]], i8* [[ATOMIC_TEMP]], align 1 272 // CHECK1-NEXT: [[TMP96:%.*]] = load i8, i8* [[ATOMIC_TEMP]], align 1 273 // CHECK1-NEXT: [[TMP97:%.*]] = cmpxchg i8* [[OMP_ARRAYCPY_DESTELEMENTPAST22]], i8 [[TMP93]], i8 [[TMP96]] monotonic monotonic, align 1 274 // CHECK1-NEXT: [[TMP98]] = extractvalue { i8, i1 } [[TMP97]], 0 275 // CHECK1-NEXT: [[TMP99:%.*]] = extractvalue { i8, i1 } [[TMP97]], 1 276 // CHECK1-NEXT: br i1 [[TMP99]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]] 277 // CHECK1: atomic_exit: 278 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT29]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST22]], i32 1 279 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT30]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST21]], i32 1 280 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE31:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT29]], [[TMP91]] 281 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE31]], label [[OMP_ARRAYCPY_DONE32]], label [[OMP_ARRAYCPY_BODY20]] 282 // CHECK1: omp.arraycpy.done32: 283 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 284 // CHECK1: .omp.reduction.default: 285 // CHECK1-NEXT: [[TMP100:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 286 // CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP100]]) 287 // CHECK1-NEXT: ret void 288 // 289 // 290 // CHECK1-LABEL: define {{[^@]+}}@.red_init. 291 // CHECK1-SAME: (i8* noalias noundef [[TMP0:%.*]], i8* noalias noundef [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { 292 // CHECK1-NEXT: entry: 293 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 294 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 295 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 296 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 297 // CHECK1-NEXT: [[TMP2:%.*]] = bitcast i8** [[DOTADDR]] to i32** 298 // CHECK1-NEXT: [[TMP3:%.*]] = load i32*, i32** [[TMP2]], align 8 299 // CHECK1-NEXT: store i32 0, i32* [[TMP3]], align 4 300 // CHECK1-NEXT: ret void 301 // 302 // 303 // CHECK1-LABEL: define {{[^@]+}}@.red_comb. 304 // CHECK1-SAME: (i8* noundef [[TMP0:%.*]], i8* noundef [[TMP1:%.*]]) #[[ATTR3]] { 305 // CHECK1-NEXT: entry: 306 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 307 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 308 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 309 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 310 // CHECK1-NEXT: [[TMP2:%.*]] = bitcast i8** [[DOTADDR]] to i32** 311 // CHECK1-NEXT: [[TMP3:%.*]] = load i32*, i32** [[TMP2]], align 8 312 // CHECK1-NEXT: [[TMP4:%.*]] = bitcast i8** [[DOTADDR1]] to i32** 313 // CHECK1-NEXT: [[TMP5:%.*]] = load i32*, i32** [[TMP4]], align 8 314 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP3]], align 4 315 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP5]], align 4 316 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP6]], [[TMP7]] 317 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP3]], align 4 318 // CHECK1-NEXT: ret void 319 // 320 // 321 // CHECK1-LABEL: define {{[^@]+}}@.red_init..1 322 // CHECK1-SAME: (i8* noalias noundef [[TMP0:%.*]], i8* noalias noundef [[TMP1:%.*]]) #[[ATTR3]] { 323 // CHECK1-NEXT: entry: 324 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 325 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 326 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 327 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 328 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 329 // CHECK1-NEXT: [[TMP3:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8 330 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr i8, i8* [[TMP2]], i64 [[TMP3]] 331 // CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[TMP2]], [[TMP4]] 332 // CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 333 // CHECK1: omp.arrayinit.body: 334 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP2]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 335 // CHECK1-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 336 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 337 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]] 338 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 339 // CHECK1: omp.arrayinit.done: 340 // CHECK1-NEXT: ret void 341 // 342 // 343 // CHECK1-LABEL: define {{[^@]+}}@.red_comb..2 344 // CHECK1-SAME: (i8* noundef [[TMP0:%.*]], i8* noundef [[TMP1:%.*]]) #[[ATTR3]] { 345 // CHECK1-NEXT: entry: 346 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 347 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 348 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 349 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 350 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8 351 // CHECK1-NEXT: [[TMP3:%.*]] = load i8*, i8** [[DOTADDR]], align 8 352 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 353 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr i8, i8* [[TMP3]], i64 [[TMP2]] 354 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP3]], [[TMP5]] 355 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 356 // CHECK1: omp.arraycpy.body: 357 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP4]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 358 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP3]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 359 // CHECK1-NEXT: [[TMP6:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 360 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP6]] to i32 361 // CHECK1-NEXT: [[TMP7:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 362 // CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP7]] to i32 363 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 364 // CHECK1-NEXT: [[CONV3:%.*]] = trunc i32 [[ADD]] to i8 365 // CHECK1-NEXT: store i8 [[CONV3]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 366 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 367 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 368 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP5]] 369 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]] 370 // CHECK1: omp.arraycpy.done4: 371 // CHECK1-NEXT: ret void 372 // 373 // 374 // CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map. 375 // CHECK1-SAME: (%struct..kmp_privates.t* noalias noundef [[TMP0:%.*]], i8*** noalias noundef [[TMP1:%.*]]) #[[ATTR6:[0-9]+]] { 376 // CHECK1-NEXT: entry: 377 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca %struct..kmp_privates.t*, align 8 378 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8***, align 8 379 // CHECK1-NEXT: store %struct..kmp_privates.t* [[TMP0]], %struct..kmp_privates.t** [[DOTADDR]], align 8 380 // CHECK1-NEXT: store i8*** [[TMP1]], i8**** [[DOTADDR1]], align 8 381 // CHECK1-NEXT: [[TMP2:%.*]] = load %struct..kmp_privates.t*, %struct..kmp_privates.t** [[DOTADDR]], align 8 382 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP2]], i32 0, i32 0 383 // CHECK1-NEXT: [[TMP4:%.*]] = load i8***, i8**** [[DOTADDR1]], align 8 384 // CHECK1-NEXT: store i8** [[TMP3]], i8*** [[TMP4]], align 8 385 // CHECK1-NEXT: ret void 386 // 387 // 388 // CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry. 389 // CHECK1-SAME: (i32 noundef [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias noundef [[TMP1:%.*]]) #[[ATTR3]] { 390 // CHECK1-NEXT: entry: 391 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4 392 // CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8 393 // CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8 394 // CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8 395 // CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8 396 // CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon*, align 8 397 // CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca i8**, align 8 398 // CHECK1-NEXT: [[TMP_I:%.*]] = alloca i8**, align 8 399 // CHECK1-NEXT: [[TMP4_I:%.*]] = alloca i8*, align 8 400 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4 401 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8 402 // CHECK1-NEXT: store i32 [[TMP0]], i32* [[DOTADDR]], align 4 403 // CHECK1-NEXT: store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8 404 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4 405 // CHECK1-NEXT: [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8 406 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0 407 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2 408 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0 409 // CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 410 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon* 411 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1 412 // CHECK1-NEXT: [[TMP10:%.*]] = bitcast %struct..kmp_privates.t* [[TMP9]] to i8* 413 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast %struct.kmp_task_t_with_privates* [[TMP3]] to i8* 414 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META3:![0-9]+]]) 415 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META6:![0-9]+]]) 416 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META8:![0-9]+]]) 417 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]]) 418 // CHECK1-NEXT: store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 419 // CHECK1-NEXT: store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !12 420 // CHECK1-NEXT: store i8* [[TMP10]], i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 421 // 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 !12 422 // CHECK1-NEXT: store i8* [[TMP11]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !12 423 // CHECK1-NEXT: store %struct.anon* [[TMP8]], %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !12 424 // CHECK1-NEXT: [[TMP12:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !12 425 // CHECK1-NEXT: [[TMP13:%.*]] = load void (i8*, ...)*, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12 426 // CHECK1-NEXT: [[TMP14:%.*]] = load i8*, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 427 // CHECK1-NEXT: [[TMP15:%.*]] = bitcast void (i8*, ...)* [[TMP13]] to void (i8*, i8***)* 428 // CHECK1-NEXT: call void [[TMP15]](i8* [[TMP14]], i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR5]] 429 // CHECK1-NEXT: [[TMP16:%.*]] = load i8**, i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias !12 430 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP12]], i32 0, i32 1 431 // CHECK1-NEXT: [[TMP18:%.*]] = load i32*, i32** [[TMP17]], align 8 432 // CHECK1-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP16]], align 8 433 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 434 // CHECK1-NEXT: [[TMP21:%.*]] = bitcast i32* [[TMP18]] to i8* 435 // CHECK1-NEXT: [[TMP22:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP19]], i8* [[TMP21]]) #[[ATTR5]] 436 // CHECK1-NEXT: [[CONV_I:%.*]] = bitcast i8* [[TMP22]] to i32* 437 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2 438 // CHECK1-NEXT: [[TMP24:%.*]] = load i8**, i8*** [[TMP23]], align 8 439 // CHECK1-NEXT: [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8 440 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 1 441 // CHECK1-NEXT: [[TMP27:%.*]] = load i32*, i32** [[TMP26]], align 8 442 // CHECK1-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4 443 // CHECK1-NEXT: [[TMP29:%.*]] = sext i32 [[TMP28]] to i64 444 // CHECK1-NEXT: [[LB_ADD_LEN_I:%.*]] = add nsw i64 -1, [[TMP29]] 445 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2 446 // CHECK1-NEXT: [[TMP31:%.*]] = load i8**, i8*** [[TMP30]], align 8 447 // CHECK1-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds i8*, i8** [[TMP31]], i64 9 448 // CHECK1-NEXT: [[TMP32:%.*]] = load i8*, i8** [[ARRAYIDX2_I]], align 8 449 // CHECK1-NEXT: [[ARRAYIDX3_I:%.*]] = getelementptr inbounds i8, i8* [[TMP32]], i64 [[LB_ADD_LEN_I]] 450 // CHECK1-NEXT: [[TMP33:%.*]] = ptrtoint i8* [[ARRAYIDX3_I]] to i64 451 // CHECK1-NEXT: [[TMP34:%.*]] = ptrtoint i8* [[TMP25]] to i64 452 // CHECK1-NEXT: [[TMP35:%.*]] = sub i64 [[TMP33]], [[TMP34]] 453 // CHECK1-NEXT: [[TMP36:%.*]] = sdiv exact i64 [[TMP35]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 454 // CHECK1-NEXT: [[TMP37:%.*]] = add nuw i64 [[TMP36]], 1 455 // CHECK1-NEXT: [[TMP38:%.*]] = mul nuw i64 [[TMP37]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 456 // CHECK1-NEXT: store i64 [[TMP37]], i64* @{{reduction_size[.].+[.]}}, align 8, !noalias !12 457 // CHECK1-NEXT: [[TMP39:%.*]] = load i8*, i8** [[TMP16]], align 8 458 // CHECK1-NEXT: [[TMP40:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP39]], i8* [[TMP25]]) #[[ATTR5]] 459 // CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2 460 // CHECK1-NEXT: [[TMP42:%.*]] = load i8**, i8*** [[TMP41]], align 8 461 // CHECK1-NEXT: [[TMP43:%.*]] = load i8*, i8** [[TMP42]], align 8 462 // CHECK1-NEXT: [[TMP44:%.*]] = ptrtoint i8* [[TMP43]] to i64 463 // CHECK1-NEXT: [[TMP45:%.*]] = ptrtoint i8* [[TMP25]] to i64 464 // CHECK1-NEXT: [[TMP46:%.*]] = sub i64 [[TMP44]], [[TMP45]] 465 // CHECK1-NEXT: [[TMP47:%.*]] = sdiv exact i64 [[TMP46]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 466 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr i8, i8* [[TMP40]], i64 [[TMP47]] 467 // CHECK1-NEXT: store i8** [[TMP4_I]], i8*** [[TMP_I]], align 8, !noalias !12 468 // CHECK1-NEXT: store i8* [[TMP48]], i8** [[TMP4_I]], align 8, !noalias !12 469 // CHECK1-NEXT: ret i32 0 470 // 471 // 472 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func 473 // CHECK1-SAME: (i8* noundef [[TMP0:%.*]], i8* noundef [[TMP1:%.*]]) #[[ATTR3]] { 474 // CHECK1-NEXT: entry: 475 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 476 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 477 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 478 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 479 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 480 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [3 x i8*]* 481 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 482 // CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [3 x i8*]* 483 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 0 484 // CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 485 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 486 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 0 487 // CHECK1-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 488 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 489 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 1 490 // CHECK1-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 491 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1 492 // CHECK1-NEXT: [[TMP15:%.*]] = load i8*, i8** [[TMP14]], align 8 493 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2 494 // CHECK1-NEXT: [[TMP17:%.*]] = load i8*, i8** [[TMP16]], align 8 495 // CHECK1-NEXT: [[TMP18:%.*]] = ptrtoint i8* [[TMP17]] to i64 496 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP11]], align 4 497 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP8]], align 4 498 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 499 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4 500 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, i8* [[TMP15]], i64 [[TMP18]] 501 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP15]], [[TMP21]] 502 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 503 // CHECK1: omp.arraycpy.body: 504 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP13]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 505 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP15]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 506 // CHECK1-NEXT: [[TMP22:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 507 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP22]] to i32 508 // CHECK1-NEXT: [[TMP23:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 509 // CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP23]] to i32 510 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 511 // CHECK1-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8 512 // CHECK1-NEXT: store i8 [[CONV4]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 513 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 514 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 515 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP21]] 516 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]] 517 // CHECK1: omp.arraycpy.done5: 518 // CHECK1-NEXT: ret void 519 // 520