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 -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 -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 -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=CHECK2 5 6 // RUN: %clang_cc1 -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 -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 -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 parallel 15 #pragma omp sections reduction(task, +: argc, argv[0:10][0:argc]) 16 { 17 #pragma omp task in_reduction(+: argc, argv[0:10][0:argc]) 18 ; 19 } 20 } 21 22 23 24 // Init firstprivate copy of argc 25 26 // Init firstprivate copy of argv[0:10][0:argc] 27 28 // Register task reduction. 29 30 31 32 33 34 35 36 37 38 #endif 39 // CHECK1-LABEL: define {{[^@]+}}@main 40 // CHECK1-SAME: (i32 [[ARGC:%.*]], i8** [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 41 // CHECK1-NEXT: entry: 42 // CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 43 // CHECK1-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 44 // CHECK1-NEXT: [[OMP_OUTLINED_ARG_AGG_:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8 45 // CHECK1-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 46 // CHECK1-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 47 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[OMP_OUTLINED_ARG_AGG_]], i32 0, i32 0 48 // CHECK1-NEXT: store i32* [[ARGC_ADDR]], i32** [[TMP0]], align 8 49 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[OMP_OUTLINED_ARG_AGG_]], i32 0, i32 1 50 // CHECK1-NEXT: store i8*** [[ARGV_ADDR]], 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_SECTIONS_LB_:%.*]] = alloca i32, align 4 62 // CHECK1-NEXT: [[DOTOMP_SECTIONS_UB_:%.*]] = alloca i32, align 4 63 // CHECK1-NEXT: [[DOTOMP_SECTIONS_ST_:%.*]] = alloca i32, align 4 64 // CHECK1-NEXT: [[DOTOMP_SECTIONS_IL_:%.*]] = alloca i32, align 4 65 // CHECK1-NEXT: [[DOTOMP_SECTIONS_IV_:%.*]] = alloca i32, align 4 66 // CHECK1-NEXT: [[ARGC:%.*]] = alloca i32, align 4 67 // CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 68 // CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 69 // CHECK1-NEXT: [[TMP:%.*]] = alloca i8**, align 8 70 // CHECK1-NEXT: [[_TMP4:%.*]] = alloca i8*, align 8 71 // CHECK1-NEXT: [[DOTRD_INPUT_:%.*]] = alloca [2 x %struct.kmp_taskred_input_t], align 8 72 // CHECK1-NEXT: [[DOTTASK_RED_:%.*]] = alloca i8*, align 8 73 // CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_0:%.*]], align 8 74 // CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8 75 // CHECK1-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1 76 // CHECK1-NEXT: [[_TMP23:%.*]] = alloca i8, align 1 77 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 78 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 79 // CHECK1-NEXT: store %struct.anon* [[__CONTEXT]], %struct.anon** [[__CONTEXT_ADDR]], align 8 80 // CHECK1-NEXT: [[TMP0:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR]], align 8 81 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP0]], i32 0, i32 0 82 // CHECK1-NEXT: [[TMP2:%.*]] = load i32*, i32** [[TMP1]], align 8 83 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP0]], i32 0, i32 1 84 // CHECK1-NEXT: [[TMP4:%.*]] = load i8***, i8**** [[TMP3]], align 8 85 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_SECTIONS_LB_]], align 4 86 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_SECTIONS_UB_]], align 4 87 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_SECTIONS_ST_]], align 4 88 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_SECTIONS_IL_]], align 4 89 // CHECK1-NEXT: store i32 0, i32* [[ARGC]], align 4 90 // CHECK1-NEXT: [[TMP5:%.*]] = load i8**, i8*** [[TMP4]], align 8 91 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8*, i8** [[TMP5]], i64 0 92 // CHECK1-NEXT: [[TMP6:%.*]] = load i8*, i8** [[ARRAYIDX]], align 8 93 // CHECK1-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, i8* [[TMP6]], i64 0 94 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP2]], align 4 95 // CHECK1-NEXT: [[TMP8:%.*]] = sext i32 [[TMP7]] to i64 96 // CHECK1-NEXT: [[LB_ADD_LEN:%.*]] = add nsw i64 -1, [[TMP8]] 97 // CHECK1-NEXT: [[TMP9:%.*]] = load i8**, i8*** [[TMP4]], align 8 98 // CHECK1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8*, i8** [[TMP9]], i64 9 99 // CHECK1-NEXT: [[TMP10:%.*]] = load i8*, i8** [[ARRAYIDX2]], align 8 100 // CHECK1-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds i8, i8* [[TMP10]], i64 [[LB_ADD_LEN]] 101 // CHECK1-NEXT: [[TMP11:%.*]] = ptrtoint i8* [[ARRAYIDX3]] to i64 102 // CHECK1-NEXT: [[TMP12:%.*]] = ptrtoint i8* [[ARRAYIDX1]] to i64 103 // CHECK1-NEXT: [[TMP13:%.*]] = sub i64 [[TMP11]], [[TMP12]] 104 // CHECK1-NEXT: [[TMP14:%.*]] = sdiv exact i64 [[TMP13]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 105 // CHECK1-NEXT: [[TMP15:%.*]] = add nuw i64 [[TMP14]], 1 106 // CHECK1-NEXT: [[TMP16:%.*]] = mul nuw i64 [[TMP15]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 107 // CHECK1-NEXT: [[TMP17:%.*]] = call i8* @llvm.stacksave() 108 // CHECK1-NEXT: store i8* [[TMP17]], i8** [[SAVED_STACK]], align 8 109 // CHECK1-NEXT: [[VLA:%.*]] = alloca i8, i64 [[TMP15]], align 16 110 // CHECK1-NEXT: store i64 [[TMP15]], i64* [[__VLA_EXPR0]], align 8 111 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP15]] 112 // CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[VLA]], [[TMP18]] 113 // CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 114 // CHECK1: omp.arrayinit.body: 115 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 116 // CHECK1-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 117 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 118 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP18]] 119 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 120 // CHECK1: omp.arrayinit.done: 121 // CHECK1-NEXT: [[TMP19:%.*]] = load i8**, i8*** [[TMP4]], align 8 122 // CHECK1-NEXT: [[TMP20:%.*]] = load i8*, i8** [[TMP19]], align 8 123 // CHECK1-NEXT: [[TMP21:%.*]] = ptrtoint i8* [[TMP20]] to i64 124 // CHECK1-NEXT: [[TMP22:%.*]] = ptrtoint i8* [[ARRAYIDX1]] to i64 125 // CHECK1-NEXT: [[TMP23:%.*]] = sub i64 [[TMP21]], [[TMP22]] 126 // CHECK1-NEXT: [[TMP24:%.*]] = sdiv exact i64 [[TMP23]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 127 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP24]] 128 // CHECK1-NEXT: store i8** [[_TMP4]], i8*** [[TMP]], align 8 129 // CHECK1-NEXT: store i8* [[TMP25]], i8** [[_TMP4]], align 8 130 // 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 0 131 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T:%.*]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 0 132 // CHECK1-NEXT: [[TMP27:%.*]] = bitcast i32* [[ARGC]] to i8* 133 // CHECK1-NEXT: store i8* [[TMP27]], i8** [[TMP26]], align 8 134 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 1 135 // CHECK1-NEXT: [[TMP29:%.*]] = bitcast i32* [[TMP2]] to i8* 136 // CHECK1-NEXT: store i8* [[TMP29]], i8** [[TMP28]], align 8 137 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 2 138 // CHECK1-NEXT: store i64 4, i64* [[TMP30]], align 8 139 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 3 140 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init. to i8*), i8** [[TMP31]], align 8 141 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 4 142 // CHECK1-NEXT: store i8* null, i8** [[TMP32]], align 8 143 // CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 5 144 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb. to i8*), i8** [[TMP33]], align 8 145 // CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 6 146 // CHECK1-NEXT: [[TMP35:%.*]] = bitcast i32* [[TMP34]] to i8* 147 // CHECK1-NEXT: call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP35]], i8 0, i64 4, i1 false) 148 // CHECK1-NEXT: [[DOTRD_INPUT_GEP_5:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]], i64 0, i64 1 149 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_5]], i32 0, i32 0 150 // CHECK1-NEXT: [[TMP37:%.*]] = load i8**, i8*** [[TMP4]], align 8 151 // CHECK1-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i8*, i8** [[TMP37]], i64 0 152 // CHECK1-NEXT: [[TMP38:%.*]] = load i8*, i8** [[ARRAYIDX6]], align 8 153 // CHECK1-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds i8, i8* [[TMP38]], i64 0 154 // CHECK1-NEXT: [[TMP39:%.*]] = load i32, i32* [[TMP2]], align 4 155 // CHECK1-NEXT: [[TMP40:%.*]] = sext i32 [[TMP39]] to i64 156 // CHECK1-NEXT: [[LB_ADD_LEN8:%.*]] = add nsw i64 -1, [[TMP40]] 157 // CHECK1-NEXT: [[TMP41:%.*]] = load i8**, i8*** [[TMP4]], align 8 158 // CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds i8*, i8** [[TMP41]], i64 9 159 // CHECK1-NEXT: [[TMP42:%.*]] = load i8*, i8** [[ARRAYIDX9]], align 8 160 // CHECK1-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i8, i8* [[TMP42]], i64 [[LB_ADD_LEN8]] 161 // CHECK1-NEXT: store i8* [[VLA]], i8** [[TMP36]], align 8 162 // CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_5]], i32 0, i32 1 163 // CHECK1-NEXT: store i8* [[ARRAYIDX7]], i8** [[TMP43]], align 8 164 // CHECK1-NEXT: [[TMP44:%.*]] = ptrtoint i8* [[ARRAYIDX10]] to i64 165 // CHECK1-NEXT: [[TMP45:%.*]] = ptrtoint i8* [[ARRAYIDX7]] to i64 166 // CHECK1-NEXT: [[TMP46:%.*]] = sub i64 [[TMP44]], [[TMP45]] 167 // CHECK1-NEXT: [[TMP47:%.*]] = sdiv exact i64 [[TMP46]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 168 // CHECK1-NEXT: [[TMP48:%.*]] = add nuw i64 [[TMP47]], 1 169 // CHECK1-NEXT: [[TMP49:%.*]] = mul nuw i64 [[TMP48]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 170 // CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_5]], i32 0, i32 2 171 // CHECK1-NEXT: store i64 [[TMP49]], i64* [[TMP50]], align 8 172 // CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_5]], i32 0, i32 3 173 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init..1 to i8*), i8** [[TMP51]], align 8 174 // CHECK1-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_5]], i32 0, i32 4 175 // CHECK1-NEXT: store i8* null, i8** [[TMP52]], align 8 176 // CHECK1-NEXT: [[TMP53:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_5]], i32 0, i32 5 177 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb..2 to i8*), i8** [[TMP53]], align 8 178 // CHECK1-NEXT: [[TMP54:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_5]], i32 0, i32 6 179 // CHECK1-NEXT: store i32 1, i32* [[TMP54]], align 8 180 // CHECK1-NEXT: [[TMP55:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 181 // CHECK1-NEXT: [[TMP56:%.*]] = load i32, i32* [[TMP55]], align 4 182 // CHECK1-NEXT: [[TMP57:%.*]] = bitcast [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]] to i8* 183 // CHECK1-NEXT: [[TMP58:%.*]] = call i8* @__kmpc_taskred_modifier_init(%struct.ident_t* @[[GLOB1]], i32 [[TMP56]], i32 1, i32 2, i8* [[TMP57]]) 184 // CHECK1-NEXT: store i8* [[TMP58]], i8** [[DOTTASK_RED_]], align 8 185 // CHECK1-NEXT: [[TMP59:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 186 // CHECK1-NEXT: [[TMP60:%.*]] = load i32, i32* [[TMP59]], align 4 187 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP60]], i32 34, i32* [[DOTOMP_SECTIONS_IL_]], i32* [[DOTOMP_SECTIONS_LB_]], i32* [[DOTOMP_SECTIONS_UB_]], i32* [[DOTOMP_SECTIONS_ST_]], i32 1, i32 1) 188 // CHECK1-NEXT: [[TMP61:%.*]] = load i32, i32* [[DOTOMP_SECTIONS_UB_]], align 4 189 // CHECK1-NEXT: [[TMP62:%.*]] = icmp slt i32 [[TMP61]], 0 190 // CHECK1-NEXT: [[TMP63:%.*]] = select i1 [[TMP62]], i32 [[TMP61]], i32 0 191 // CHECK1-NEXT: store i32 [[TMP63]], i32* [[DOTOMP_SECTIONS_UB_]], align 4 192 // CHECK1-NEXT: [[TMP64:%.*]] = load i32, i32* [[DOTOMP_SECTIONS_LB_]], align 4 193 // CHECK1-NEXT: store i32 [[TMP64]], i32* [[DOTOMP_SECTIONS_IV_]], align 4 194 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 195 // CHECK1: omp.inner.for.cond: 196 // CHECK1-NEXT: [[TMP65:%.*]] = load i32, i32* [[DOTOMP_SECTIONS_IV_]], align 4 197 // CHECK1-NEXT: [[TMP66:%.*]] = load i32, i32* [[DOTOMP_SECTIONS_UB_]], align 4 198 // CHECK1-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP65]], [[TMP66]] 199 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 200 // CHECK1: omp.inner.for.body: 201 // CHECK1-NEXT: [[TMP67:%.*]] = load i32, i32* [[DOTOMP_SECTIONS_IV_]], align 4 202 // CHECK1-NEXT: switch i32 [[TMP67]], label [[DOTOMP_SECTIONS_EXIT:%.*]] [ 203 // CHECK1-NEXT: i32 0, label [[DOTOMP_SECTIONS_CASE:%.*]] 204 // CHECK1-NEXT: ] 205 // CHECK1: .omp.sections.case: 206 // CHECK1-NEXT: [[TMP68:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[AGG_CAPTURED]], i32 0, i32 0 207 // CHECK1-NEXT: store i8** [[DOTTASK_RED_]], i8*** [[TMP68]], align 8 208 // CHECK1-NEXT: [[TMP69:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[AGG_CAPTURED]], i32 0, i32 1 209 // CHECK1-NEXT: store i32* [[ARGC]], i32** [[TMP69]], align 8 210 // CHECK1-NEXT: [[TMP70:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[AGG_CAPTURED]], i32 0, i32 2 211 // CHECK1-NEXT: [[TMP71:%.*]] = load i8**, i8*** [[TMP]], align 8 212 // CHECK1-NEXT: store i8** [[TMP71]], i8*** [[TMP70]], align 8 213 // CHECK1-NEXT: [[TMP72:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 214 // CHECK1-NEXT: [[TMP73:%.*]] = load i32, i32* [[TMP72]], align 4 215 // CHECK1-NEXT: [[TMP74:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP73]], 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*)*)) 216 // CHECK1-NEXT: [[TMP75:%.*]] = bitcast i8* [[TMP74]] to %struct.kmp_task_t_with_privates* 217 // CHECK1-NEXT: [[TMP76:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP75]], i32 0, i32 0 218 // CHECK1-NEXT: [[TMP77:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP76]], i32 0, i32 0 219 // CHECK1-NEXT: [[TMP78:%.*]] = load i8*, i8** [[TMP77]], align 8 220 // CHECK1-NEXT: [[TMP79:%.*]] = bitcast %struct.anon.0* [[AGG_CAPTURED]] to i8* 221 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP78]], i8* align 8 [[TMP79]], i64 24, i1 false) 222 // CHECK1-NEXT: [[TMP80:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP75]], i32 0, i32 1 223 // CHECK1-NEXT: [[TMP81:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP80]], i32 0, i32 0 224 // CHECK1-NEXT: [[TMP82:%.*]] = load i8*, i8** [[DOTTASK_RED_]], align 8 225 // CHECK1-NEXT: store i8* [[TMP82]], i8** [[TMP81]], align 8 226 // CHECK1-NEXT: [[TMP83:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 227 // CHECK1-NEXT: [[TMP84:%.*]] = load i32, i32* [[TMP83]], align 4 228 // CHECK1-NEXT: [[TMP85:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP84]], i8* [[TMP74]]) 229 // CHECK1-NEXT: br label [[DOTOMP_SECTIONS_EXIT]] 230 // CHECK1: .omp.sections.exit: 231 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 232 // CHECK1: omp.inner.for.inc: 233 // CHECK1-NEXT: [[TMP86:%.*]] = load i32, i32* [[DOTOMP_SECTIONS_IV_]], align 4 234 // CHECK1-NEXT: [[INC:%.*]] = add nsw i32 [[TMP86]], 1 235 // CHECK1-NEXT: store i32 [[INC]], i32* [[DOTOMP_SECTIONS_IV_]], align 4 236 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 237 // CHECK1: omp.inner.for.end: 238 // CHECK1-NEXT: [[TMP87:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 239 // CHECK1-NEXT: [[TMP88:%.*]] = load i32, i32* [[TMP87]], align 4 240 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP88]]) 241 // CHECK1-NEXT: [[TMP89:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 242 // CHECK1-NEXT: [[TMP90:%.*]] = load i32, i32* [[TMP89]], align 4 243 // CHECK1-NEXT: call void @__kmpc_task_reduction_modifier_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP90]], i32 1) 244 // CHECK1-NEXT: [[TMP91:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 245 // CHECK1-NEXT: [[TMP92:%.*]] = bitcast i32* [[ARGC]] to i8* 246 // CHECK1-NEXT: store i8* [[TMP92]], i8** [[TMP91]], align 8 247 // CHECK1-NEXT: [[TMP93:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 248 // CHECK1-NEXT: store i8* [[VLA]], i8** [[TMP93]], align 8 249 // CHECK1-NEXT: [[TMP94:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 250 // CHECK1-NEXT: [[TMP95:%.*]] = inttoptr i64 [[TMP15]] to i8* 251 // CHECK1-NEXT: store i8* [[TMP95]], i8** [[TMP94]], align 8 252 // CHECK1-NEXT: [[TMP96:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 253 // CHECK1-NEXT: [[TMP97:%.*]] = load i32, i32* [[TMP96]], align 4 254 // CHECK1-NEXT: [[TMP98:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 255 // CHECK1-NEXT: [[TMP99:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[TMP97]], i32 2, i64 24, i8* [[TMP98]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var) 256 // CHECK1-NEXT: switch i32 [[TMP99]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 257 // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 258 // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 259 // CHECK1-NEXT: ] 260 // CHECK1: .omp.reduction.case1: 261 // CHECK1-NEXT: [[TMP100:%.*]] = load i32, i32* [[TMP2]], align 4 262 // CHECK1-NEXT: [[TMP101:%.*]] = load i32, i32* [[ARGC]], align 4 263 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP100]], [[TMP101]] 264 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP2]], align 4 265 // CHECK1-NEXT: [[TMP102:%.*]] = getelementptr i8, i8* [[ARRAYIDX1]], i64 [[TMP15]] 266 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[ARRAYIDX1]], [[TMP102]] 267 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE17:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 268 // CHECK1: omp.arraycpy.body: 269 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 270 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST11:%.*]] = phi i8* [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT15:%.*]], [[OMP_ARRAYCPY_BODY]] ] 271 // CHECK1-NEXT: [[TMP103:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], align 1 272 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP103]] to i32 273 // CHECK1-NEXT: [[TMP104:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 274 // CHECK1-NEXT: [[CONV12:%.*]] = sext i8 [[TMP104]] to i32 275 // CHECK1-NEXT: [[ADD13:%.*]] = add nsw i32 [[CONV]], [[CONV12]] 276 // CHECK1-NEXT: [[CONV14:%.*]] = trunc i32 [[ADD13]] to i8 277 // CHECK1-NEXT: store i8 [[CONV14]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], align 1 278 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT15]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], i32 1 279 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 280 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE16:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT15]], [[TMP102]] 281 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE16]], label [[OMP_ARRAYCPY_DONE17]], label [[OMP_ARRAYCPY_BODY]] 282 // CHECK1: omp.arraycpy.done17: 283 // CHECK1-NEXT: call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP97]], [8 x i32]* @.gomp_critical_user_.reduction.var) 284 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 285 // CHECK1: .omp.reduction.case2: 286 // CHECK1-NEXT: [[TMP105:%.*]] = load i32, i32* [[ARGC]], align 4 287 // CHECK1-NEXT: [[TMP106:%.*]] = atomicrmw add i32* [[TMP2]], i32 [[TMP105]] monotonic, align 4 288 // CHECK1-NEXT: [[TMP107:%.*]] = getelementptr i8, i8* [[ARRAYIDX1]], i64 [[TMP15]] 289 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY18:%.*]] = icmp eq i8* [[ARRAYIDX1]], [[TMP107]] 290 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY18]], label [[OMP_ARRAYCPY_DONE31:%.*]], label [[OMP_ARRAYCPY_BODY19:%.*]] 291 // CHECK1: omp.arraycpy.body19: 292 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST20:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT29:%.*]], [[ATOMIC_EXIT:%.*]] ] 293 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST21:%.*]] = phi i8* [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT28:%.*]], [[ATOMIC_EXIT]] ] 294 // CHECK1-NEXT: [[TMP108:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST20]], align 1 295 // CHECK1-NEXT: [[CONV22:%.*]] = sext i8 [[TMP108]] to i32 296 // CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST21]] monotonic, align 1 297 // CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]] 298 // CHECK1: atomic_cont: 299 // CHECK1-NEXT: [[TMP109:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY19]] ], [ [[TMP114:%.*]], [[ATOMIC_CONT]] ] 300 // CHECK1-NEXT: store i8 [[TMP109]], i8* [[_TMP23]], align 1 301 // CHECK1-NEXT: [[TMP110:%.*]] = load i8, i8* [[_TMP23]], align 1 302 // CHECK1-NEXT: [[CONV24:%.*]] = sext i8 [[TMP110]] to i32 303 // CHECK1-NEXT: [[TMP111:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST20]], align 1 304 // CHECK1-NEXT: [[CONV25:%.*]] = sext i8 [[TMP111]] to i32 305 // CHECK1-NEXT: [[ADD26:%.*]] = add nsw i32 [[CONV24]], [[CONV25]] 306 // CHECK1-NEXT: [[CONV27:%.*]] = trunc i32 [[ADD26]] to i8 307 // CHECK1-NEXT: store i8 [[CONV27]], i8* [[ATOMIC_TEMP]], align 1 308 // CHECK1-NEXT: [[TMP112:%.*]] = load i8, i8* [[ATOMIC_TEMP]], align 1 309 // CHECK1-NEXT: [[TMP113:%.*]] = cmpxchg i8* [[OMP_ARRAYCPY_DESTELEMENTPAST21]], i8 [[TMP109]], i8 [[TMP112]] monotonic monotonic, align 1 310 // CHECK1-NEXT: [[TMP114]] = extractvalue { i8, i1 } [[TMP113]], 0 311 // CHECK1-NEXT: [[TMP115:%.*]] = extractvalue { i8, i1 } [[TMP113]], 1 312 // CHECK1-NEXT: br i1 [[TMP115]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]] 313 // CHECK1: atomic_exit: 314 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT28]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST21]], i32 1 315 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT29]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST20]], i32 1 316 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE30:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT28]], [[TMP107]] 317 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE30]], label [[OMP_ARRAYCPY_DONE31]], label [[OMP_ARRAYCPY_BODY19]] 318 // CHECK1: omp.arraycpy.done31: 319 // CHECK1-NEXT: call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP97]], [8 x i32]* @.gomp_critical_user_.reduction.var) 320 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 321 // CHECK1: .omp.reduction.default: 322 // CHECK1-NEXT: [[TMP116:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 323 // CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP116]]) 324 // CHECK1-NEXT: [[TMP117:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 325 // CHECK1-NEXT: [[TMP118:%.*]] = load i32, i32* [[TMP117]], align 4 326 // CHECK1-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4:[0-9]+]], i32 [[TMP118]]) 327 // CHECK1-NEXT: ret void 328 // 329 // 330 // CHECK1-LABEL: define {{[^@]+}}@.red_init. 331 // CHECK1-SAME: (i8* noalias [[TMP0:%.*]], i8* noalias [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { 332 // CHECK1-NEXT: entry: 333 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 334 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 335 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 336 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 337 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 338 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to i32* 339 // CHECK1-NEXT: store i32 0, i32* [[TMP3]], align 8 340 // CHECK1-NEXT: ret void 341 // 342 // 343 // CHECK1-LABEL: define {{[^@]+}}@.red_comb. 344 // CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[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 i8*, i8** [[DOTADDR]], align 8 351 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to i32* 352 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 353 // CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to i32* 354 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP3]], align 8 355 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP5]], align 8 356 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP6]], [[TMP7]] 357 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP3]], align 8 358 // CHECK1-NEXT: ret void 359 // 360 // 361 // CHECK1-LABEL: define {{[^@]+}}@.red_init..1 362 // CHECK1-SAME: (i8* noalias [[TMP0:%.*]], i8* noalias [[TMP1:%.*]]) #[[ATTR3]] { 363 // CHECK1-NEXT: entry: 364 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 365 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 366 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 367 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 368 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 369 // CHECK1-NEXT: [[TMP3:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8 370 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr i8, i8* [[TMP2]], i64 [[TMP3]] 371 // CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[TMP2]], [[TMP4]] 372 // CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 373 // CHECK1: omp.arrayinit.body: 374 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP2]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 375 // CHECK1-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 376 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 377 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]] 378 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 379 // CHECK1: omp.arrayinit.done: 380 // CHECK1-NEXT: ret void 381 // 382 // 383 // CHECK1-LABEL: define {{[^@]+}}@.red_comb..2 384 // CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 385 // CHECK1-NEXT: entry: 386 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 387 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 388 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 389 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 390 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8 391 // CHECK1-NEXT: [[TMP3:%.*]] = load i8*, i8** [[DOTADDR]], align 8 392 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 393 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr i8, i8* [[TMP3]], i64 [[TMP2]] 394 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP3]], [[TMP5]] 395 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 396 // CHECK1: omp.arraycpy.body: 397 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP4]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 398 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP3]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 399 // CHECK1-NEXT: [[TMP6:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 400 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP6]] to i32 401 // CHECK1-NEXT: [[TMP7:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 402 // CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP7]] to i32 403 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 404 // CHECK1-NEXT: [[CONV3:%.*]] = trunc i32 [[ADD]] to i8 405 // CHECK1-NEXT: store i8 [[CONV3]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 406 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 407 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 408 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP5]] 409 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]] 410 // CHECK1: omp.arraycpy.done4: 411 // CHECK1-NEXT: ret void 412 // 413 // 414 // CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map. 415 // CHECK1-SAME: (%struct..kmp_privates.t* noalias [[TMP0:%.*]], i8*** noalias [[TMP1:%.*]]) #[[ATTR6:[0-9]+]] { 416 // CHECK1-NEXT: entry: 417 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca %struct..kmp_privates.t*, align 8 418 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8***, align 8 419 // CHECK1-NEXT: store %struct..kmp_privates.t* [[TMP0]], %struct..kmp_privates.t** [[DOTADDR]], align 8 420 // CHECK1-NEXT: store i8*** [[TMP1]], i8**** [[DOTADDR1]], align 8 421 // CHECK1-NEXT: [[TMP2:%.*]] = load %struct..kmp_privates.t*, %struct..kmp_privates.t** [[DOTADDR]], align 8 422 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP2]], i32 0, i32 0 423 // CHECK1-NEXT: [[TMP4:%.*]] = load i8***, i8**** [[DOTADDR1]], align 8 424 // CHECK1-NEXT: store i8** [[TMP3]], i8*** [[TMP4]], align 8 425 // CHECK1-NEXT: ret void 426 // 427 // 428 // CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry. 429 // CHECK1-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias [[TMP1:%.*]]) #[[ATTR3]] { 430 // CHECK1-NEXT: entry: 431 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4 432 // CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8 433 // CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8 434 // CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8 435 // CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8 436 // CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon.0*, align 8 437 // CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca i8**, align 8 438 // CHECK1-NEXT: [[TMP_I:%.*]] = alloca i8**, align 8 439 // CHECK1-NEXT: [[TMP4_I:%.*]] = alloca i8*, align 8 440 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4 441 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8 442 // CHECK1-NEXT: store i32 [[TMP0]], i32* [[DOTADDR]], align 4 443 // CHECK1-NEXT: store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8 444 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4 445 // CHECK1-NEXT: [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8 446 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0 447 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2 448 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0 449 // CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 450 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon.0* 451 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1 452 // CHECK1-NEXT: [[TMP10:%.*]] = bitcast %struct..kmp_privates.t* [[TMP9]] to i8* 453 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast %struct.kmp_task_t_with_privates* [[TMP3]] to i8* 454 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META3:![0-9]+]]) 455 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META6:![0-9]+]]) 456 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META8:![0-9]+]]) 457 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]]) 458 // CHECK1-NEXT: store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 459 // CHECK1-NEXT: store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !12 460 // CHECK1-NEXT: store i8* [[TMP10]], i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 461 // 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 !12 462 // CHECK1-NEXT: store i8* [[TMP11]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !12 463 // CHECK1-NEXT: store %struct.anon.0* [[TMP8]], %struct.anon.0** [[__CONTEXT_ADDR_I]], align 8, !noalias !12 464 // CHECK1-NEXT: [[TMP12:%.*]] = load %struct.anon.0*, %struct.anon.0** [[__CONTEXT_ADDR_I]], align 8, !noalias !12 465 // CHECK1-NEXT: [[TMP13:%.*]] = load void (i8*, ...)*, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12 466 // CHECK1-NEXT: [[TMP14:%.*]] = load i8*, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 467 // CHECK1-NEXT: [[TMP15:%.*]] = bitcast void (i8*, ...)* [[TMP13]] to void (i8*, i8***)* 468 // CHECK1-NEXT: call void [[TMP15]](i8* [[TMP14]], i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR5:[0-9]+]] 469 // CHECK1-NEXT: [[TMP16:%.*]] = load i8**, i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias !12 470 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_ANON_0:%.*]], %struct.anon.0* [[TMP12]], i32 0, i32 1 471 // CHECK1-NEXT: [[TMP18:%.*]] = load i32*, i32** [[TMP17]], align 8 472 // CHECK1-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP16]], align 8 473 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 474 // CHECK1-NEXT: [[TMP21:%.*]] = bitcast i32* [[TMP18]] to i8* 475 // CHECK1-NEXT: [[TMP22:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP19]], i8* [[TMP21]]) #[[ATTR5]] 476 // CHECK1-NEXT: [[CONV_I:%.*]] = bitcast i8* [[TMP22]] to i32* 477 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[TMP12]], i32 0, i32 2 478 // CHECK1-NEXT: [[TMP24:%.*]] = load i8**, i8*** [[TMP23]], align 8 479 // CHECK1-NEXT: [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8 480 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[TMP12]], i32 0, i32 1 481 // CHECK1-NEXT: [[TMP27:%.*]] = load i32*, i32** [[TMP26]], align 8 482 // CHECK1-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4 483 // CHECK1-NEXT: [[TMP29:%.*]] = sext i32 [[TMP28]] to i64 484 // CHECK1-NEXT: [[LB_ADD_LEN_I:%.*]] = add nsw i64 -1, [[TMP29]] 485 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[TMP12]], i32 0, i32 2 486 // CHECK1-NEXT: [[TMP31:%.*]] = load i8**, i8*** [[TMP30]], align 8 487 // CHECK1-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds i8*, i8** [[TMP31]], i64 9 488 // CHECK1-NEXT: [[TMP32:%.*]] = load i8*, i8** [[ARRAYIDX2_I]], align 8 489 // CHECK1-NEXT: [[ARRAYIDX3_I:%.*]] = getelementptr inbounds i8, i8* [[TMP32]], i64 [[LB_ADD_LEN_I]] 490 // CHECK1-NEXT: [[TMP33:%.*]] = ptrtoint i8* [[ARRAYIDX3_I]] to i64 491 // CHECK1-NEXT: [[TMP34:%.*]] = ptrtoint i8* [[TMP25]] to i64 492 // CHECK1-NEXT: [[TMP35:%.*]] = sub i64 [[TMP33]], [[TMP34]] 493 // CHECK1-NEXT: [[TMP36:%.*]] = sdiv exact i64 [[TMP35]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 494 // CHECK1-NEXT: [[TMP37:%.*]] = add nuw i64 [[TMP36]], 1 495 // CHECK1-NEXT: [[TMP38:%.*]] = mul nuw i64 [[TMP37]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 496 // CHECK1-NEXT: store i64 [[TMP37]], i64* @{{reduction_size[.].+[.]}}, align 8 497 // CHECK1-NEXT: [[TMP39:%.*]] = load i8*, i8** [[TMP16]], align 8 498 // CHECK1-NEXT: [[TMP40:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP39]], i8* [[TMP25]]) #[[ATTR5]] 499 // CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[TMP12]], i32 0, i32 2 500 // CHECK1-NEXT: [[TMP42:%.*]] = load i8**, i8*** [[TMP41]], align 8 501 // CHECK1-NEXT: [[TMP43:%.*]] = load i8*, i8** [[TMP42]], align 8 502 // CHECK1-NEXT: [[TMP44:%.*]] = ptrtoint i8* [[TMP43]] to i64 503 // CHECK1-NEXT: [[TMP45:%.*]] = ptrtoint i8* [[TMP25]] to i64 504 // CHECK1-NEXT: [[TMP46:%.*]] = sub i64 [[TMP44]], [[TMP45]] 505 // CHECK1-NEXT: [[TMP47:%.*]] = sdiv exact i64 [[TMP46]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 506 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr i8, i8* [[TMP40]], i64 [[TMP47]] 507 // CHECK1-NEXT: store i8** [[TMP4_I]], i8*** [[TMP_I]], align 8, !noalias !12 508 // CHECK1-NEXT: store i8* [[TMP48]], i8** [[TMP4_I]], align 8, !noalias !12 509 // CHECK1-NEXT: ret i32 0 510 // 511 // 512 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func 513 // CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 514 // CHECK1-NEXT: entry: 515 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 516 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 517 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 518 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 519 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 520 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [3 x i8*]* 521 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 522 // CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [3 x i8*]* 523 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 0 524 // CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 525 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 526 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 0 527 // CHECK1-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 528 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 529 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 1 530 // CHECK1-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 531 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1 532 // CHECK1-NEXT: [[TMP15:%.*]] = load i8*, i8** [[TMP14]], align 8 533 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2 534 // CHECK1-NEXT: [[TMP17:%.*]] = load i8*, i8** [[TMP16]], align 8 535 // CHECK1-NEXT: [[TMP18:%.*]] = ptrtoint i8* [[TMP17]] to i64 536 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP11]], align 4 537 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP8]], align 4 538 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 539 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4 540 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, i8* [[TMP15]], i64 [[TMP18]] 541 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP15]], [[TMP21]] 542 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 543 // CHECK1: omp.arraycpy.body: 544 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP13]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 545 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP15]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 546 // CHECK1-NEXT: [[TMP22:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 547 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP22]] to i32 548 // CHECK1-NEXT: [[TMP23:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 549 // CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP23]] to i32 550 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 551 // CHECK1-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8 552 // CHECK1-NEXT: store i8 [[CONV4]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 553 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 554 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 555 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP21]] 556 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]] 557 // CHECK1: omp.arraycpy.done5: 558 // CHECK1-NEXT: ret void 559 // 560 // 561 // CHECK2-LABEL: define {{[^@]+}}@main 562 // CHECK2-SAME: (i32 [[ARGC:%.*]], i8** [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 563 // CHECK2-NEXT: entry: 564 // CHECK2-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 565 // CHECK2-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 566 // CHECK2-NEXT: [[OMP_OUTLINED_ARG_AGG_:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8 567 // CHECK2-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 568 // CHECK2-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 569 // CHECK2-NEXT: [[TMP0:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[OMP_OUTLINED_ARG_AGG_]], i32 0, i32 0 570 // CHECK2-NEXT: store i32* [[ARGC_ADDR]], i32** [[TMP0]], align 8 571 // CHECK2-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[OMP_OUTLINED_ARG_AGG_]], i32 0, i32 1 572 // CHECK2-NEXT: store i8*** [[ARGV_ADDR]], i8**** [[TMP1]], align 8 573 // 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_]]) 574 // CHECK2-NEXT: ret i32 0 575 // 576 // 577 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined. 578 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], %struct.anon* noalias [[__CONTEXT:%.*]]) #[[ATTR1:[0-9]+]] { 579 // CHECK2-NEXT: entry: 580 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 581 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 582 // CHECK2-NEXT: [[__CONTEXT_ADDR:%.*]] = alloca %struct.anon*, align 8 583 // CHECK2-NEXT: [[DOTOMP_SECTIONS_LB_:%.*]] = alloca i32, align 4 584 // CHECK2-NEXT: [[DOTOMP_SECTIONS_UB_:%.*]] = alloca i32, align 4 585 // CHECK2-NEXT: [[DOTOMP_SECTIONS_ST_:%.*]] = alloca i32, align 4 586 // CHECK2-NEXT: [[DOTOMP_SECTIONS_IL_:%.*]] = alloca i32, align 4 587 // CHECK2-NEXT: [[DOTOMP_SECTIONS_IV_:%.*]] = alloca i32, align 4 588 // CHECK2-NEXT: [[ARGC:%.*]] = alloca i32, align 4 589 // CHECK2-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 590 // CHECK2-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 591 // CHECK2-NEXT: [[TMP:%.*]] = alloca i8**, align 8 592 // CHECK2-NEXT: [[_TMP4:%.*]] = alloca i8*, align 8 593 // CHECK2-NEXT: [[DOTRD_INPUT_:%.*]] = alloca [2 x %struct.kmp_taskred_input_t], align 8 594 // CHECK2-NEXT: [[DOTTASK_RED_:%.*]] = alloca i8*, align 8 595 // CHECK2-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON_0:%.*]], align 8 596 // CHECK2-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8 597 // CHECK2-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1 598 // CHECK2-NEXT: [[_TMP23:%.*]] = alloca i8, align 1 599 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 600 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 601 // CHECK2-NEXT: store %struct.anon* [[__CONTEXT]], %struct.anon** [[__CONTEXT_ADDR]], align 8 602 // CHECK2-NEXT: [[TMP0:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR]], align 8 603 // CHECK2-NEXT: [[TMP1:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP0]], i32 0, i32 0 604 // CHECK2-NEXT: [[TMP2:%.*]] = load i32*, i32** [[TMP1]], align 8 605 // CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP0]], i32 0, i32 1 606 // CHECK2-NEXT: [[TMP4:%.*]] = load i8***, i8**** [[TMP3]], align 8 607 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_SECTIONS_LB_]], align 4 608 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_SECTIONS_UB_]], align 4 609 // CHECK2-NEXT: store i32 1, i32* [[DOTOMP_SECTIONS_ST_]], align 4 610 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_SECTIONS_IL_]], align 4 611 // CHECK2-NEXT: store i32 0, i32* [[ARGC]], align 4 612 // CHECK2-NEXT: [[TMP5:%.*]] = load i8**, i8*** [[TMP4]], align 8 613 // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8*, i8** [[TMP5]], i64 0 614 // CHECK2-NEXT: [[TMP6:%.*]] = load i8*, i8** [[ARRAYIDX]], align 8 615 // CHECK2-NEXT: [[ARRAYIDX1:%.*]] = getelementptr inbounds i8, i8* [[TMP6]], i64 0 616 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP2]], align 4 617 // CHECK2-NEXT: [[TMP8:%.*]] = sext i32 [[TMP7]] to i64 618 // CHECK2-NEXT: [[LB_ADD_LEN:%.*]] = add nsw i64 -1, [[TMP8]] 619 // CHECK2-NEXT: [[TMP9:%.*]] = load i8**, i8*** [[TMP4]], align 8 620 // CHECK2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8*, i8** [[TMP9]], i64 9 621 // CHECK2-NEXT: [[TMP10:%.*]] = load i8*, i8** [[ARRAYIDX2]], align 8 622 // CHECK2-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds i8, i8* [[TMP10]], i64 [[LB_ADD_LEN]] 623 // CHECK2-NEXT: [[TMP11:%.*]] = ptrtoint i8* [[ARRAYIDX3]] to i64 624 // CHECK2-NEXT: [[TMP12:%.*]] = ptrtoint i8* [[ARRAYIDX1]] to i64 625 // CHECK2-NEXT: [[TMP13:%.*]] = sub i64 [[TMP11]], [[TMP12]] 626 // CHECK2-NEXT: [[TMP14:%.*]] = sdiv exact i64 [[TMP13]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 627 // CHECK2-NEXT: [[TMP15:%.*]] = add nuw i64 [[TMP14]], 1 628 // CHECK2-NEXT: [[TMP16:%.*]] = mul nuw i64 [[TMP15]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 629 // CHECK2-NEXT: [[TMP17:%.*]] = call i8* @llvm.stacksave() 630 // CHECK2-NEXT: store i8* [[TMP17]], i8** [[SAVED_STACK]], align 8 631 // CHECK2-NEXT: [[VLA:%.*]] = alloca i8, i64 [[TMP15]], align 16 632 // CHECK2-NEXT: store i64 [[TMP15]], i64* [[__VLA_EXPR0]], align 8 633 // CHECK2-NEXT: [[TMP18:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP15]] 634 // CHECK2-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[VLA]], [[TMP18]] 635 // CHECK2-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 636 // CHECK2: omp.arrayinit.body: 637 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 638 // CHECK2-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 639 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 640 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP18]] 641 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 642 // CHECK2: omp.arrayinit.done: 643 // CHECK2-NEXT: [[TMP19:%.*]] = load i8**, i8*** [[TMP4]], align 8 644 // CHECK2-NEXT: [[TMP20:%.*]] = load i8*, i8** [[TMP19]], align 8 645 // CHECK2-NEXT: [[TMP21:%.*]] = ptrtoint i8* [[TMP20]] to i64 646 // CHECK2-NEXT: [[TMP22:%.*]] = ptrtoint i8* [[ARRAYIDX1]] to i64 647 // CHECK2-NEXT: [[TMP23:%.*]] = sub i64 [[TMP21]], [[TMP22]] 648 // CHECK2-NEXT: [[TMP24:%.*]] = sdiv exact i64 [[TMP23]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 649 // CHECK2-NEXT: [[TMP25:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP24]] 650 // CHECK2-NEXT: store i8** [[_TMP4]], i8*** [[TMP]], align 8 651 // CHECK2-NEXT: store i8* [[TMP25]], i8** [[_TMP4]], align 8 652 // 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 0 653 // CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T:%.*]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 0 654 // CHECK2-NEXT: [[TMP27:%.*]] = bitcast i32* [[ARGC]] to i8* 655 // CHECK2-NEXT: store i8* [[TMP27]], i8** [[TMP26]], align 8 656 // CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 1 657 // CHECK2-NEXT: [[TMP29:%.*]] = bitcast i32* [[TMP2]] to i8* 658 // CHECK2-NEXT: store i8* [[TMP29]], i8** [[TMP28]], align 8 659 // CHECK2-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 2 660 // CHECK2-NEXT: store i64 4, i64* [[TMP30]], align 8 661 // CHECK2-NEXT: [[TMP31:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 3 662 // CHECK2-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init. to i8*), i8** [[TMP31]], align 8 663 // CHECK2-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 4 664 // CHECK2-NEXT: store i8* null, i8** [[TMP32]], align 8 665 // CHECK2-NEXT: [[TMP33:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 5 666 // CHECK2-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb. to i8*), i8** [[TMP33]], align 8 667 // CHECK2-NEXT: [[TMP34:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 6 668 // CHECK2-NEXT: [[TMP35:%.*]] = bitcast i32* [[TMP34]] to i8* 669 // CHECK2-NEXT: call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP35]], i8 0, i64 4, i1 false) 670 // CHECK2-NEXT: [[DOTRD_INPUT_GEP_5:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]], i64 0, i64 1 671 // CHECK2-NEXT: [[TMP36:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_5]], i32 0, i32 0 672 // CHECK2-NEXT: [[TMP37:%.*]] = load i8**, i8*** [[TMP4]], align 8 673 // CHECK2-NEXT: [[ARRAYIDX6:%.*]] = getelementptr inbounds i8*, i8** [[TMP37]], i64 0 674 // CHECK2-NEXT: [[TMP38:%.*]] = load i8*, i8** [[ARRAYIDX6]], align 8 675 // CHECK2-NEXT: [[ARRAYIDX7:%.*]] = getelementptr inbounds i8, i8* [[TMP38]], i64 0 676 // CHECK2-NEXT: [[TMP39:%.*]] = load i32, i32* [[TMP2]], align 4 677 // CHECK2-NEXT: [[TMP40:%.*]] = sext i32 [[TMP39]] to i64 678 // CHECK2-NEXT: [[LB_ADD_LEN8:%.*]] = add nsw i64 -1, [[TMP40]] 679 // CHECK2-NEXT: [[TMP41:%.*]] = load i8**, i8*** [[TMP4]], align 8 680 // CHECK2-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds i8*, i8** [[TMP41]], i64 9 681 // CHECK2-NEXT: [[TMP42:%.*]] = load i8*, i8** [[ARRAYIDX9]], align 8 682 // CHECK2-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i8, i8* [[TMP42]], i64 [[LB_ADD_LEN8]] 683 // CHECK2-NEXT: store i8* [[VLA]], i8** [[TMP36]], align 8 684 // CHECK2-NEXT: [[TMP43:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_5]], i32 0, i32 1 685 // CHECK2-NEXT: store i8* [[ARRAYIDX7]], i8** [[TMP43]], align 8 686 // CHECK2-NEXT: [[TMP44:%.*]] = ptrtoint i8* [[ARRAYIDX10]] to i64 687 // CHECK2-NEXT: [[TMP45:%.*]] = ptrtoint i8* [[ARRAYIDX7]] to i64 688 // CHECK2-NEXT: [[TMP46:%.*]] = sub i64 [[TMP44]], [[TMP45]] 689 // CHECK2-NEXT: [[TMP47:%.*]] = sdiv exact i64 [[TMP46]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 690 // CHECK2-NEXT: [[TMP48:%.*]] = add nuw i64 [[TMP47]], 1 691 // CHECK2-NEXT: [[TMP49:%.*]] = mul nuw i64 [[TMP48]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 692 // CHECK2-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_5]], i32 0, i32 2 693 // CHECK2-NEXT: store i64 [[TMP49]], i64* [[TMP50]], align 8 694 // CHECK2-NEXT: [[TMP51:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_5]], i32 0, i32 3 695 // CHECK2-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init..1 to i8*), i8** [[TMP51]], align 8 696 // CHECK2-NEXT: [[TMP52:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_5]], i32 0, i32 4 697 // CHECK2-NEXT: store i8* null, i8** [[TMP52]], align 8 698 // CHECK2-NEXT: [[TMP53:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_5]], i32 0, i32 5 699 // CHECK2-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb..2 to i8*), i8** [[TMP53]], align 8 700 // CHECK2-NEXT: [[TMP54:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_5]], i32 0, i32 6 701 // CHECK2-NEXT: store i32 1, i32* [[TMP54]], align 8 702 // CHECK2-NEXT: [[TMP55:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 703 // CHECK2-NEXT: [[TMP56:%.*]] = load i32, i32* [[TMP55]], align 4 704 // CHECK2-NEXT: [[TMP57:%.*]] = bitcast [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]] to i8* 705 // CHECK2-NEXT: [[TMP58:%.*]] = call i8* @__kmpc_taskred_modifier_init(%struct.ident_t* @[[GLOB1]], i32 [[TMP56]], i32 1, i32 2, i8* [[TMP57]]) 706 // CHECK2-NEXT: store i8* [[TMP58]], i8** [[DOTTASK_RED_]], align 8 707 // CHECK2-NEXT: [[TMP59:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 708 // CHECK2-NEXT: [[TMP60:%.*]] = load i32, i32* [[TMP59]], align 4 709 // CHECK2-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP60]], i32 34, i32* [[DOTOMP_SECTIONS_IL_]], i32* [[DOTOMP_SECTIONS_LB_]], i32* [[DOTOMP_SECTIONS_UB_]], i32* [[DOTOMP_SECTIONS_ST_]], i32 1, i32 1) 710 // CHECK2-NEXT: [[TMP61:%.*]] = load i32, i32* [[DOTOMP_SECTIONS_UB_]], align 4 711 // CHECK2-NEXT: [[TMP62:%.*]] = icmp slt i32 [[TMP61]], 0 712 // CHECK2-NEXT: [[TMP63:%.*]] = select i1 [[TMP62]], i32 [[TMP61]], i32 0 713 // CHECK2-NEXT: store i32 [[TMP63]], i32* [[DOTOMP_SECTIONS_UB_]], align 4 714 // CHECK2-NEXT: [[TMP64:%.*]] = load i32, i32* [[DOTOMP_SECTIONS_LB_]], align 4 715 // CHECK2-NEXT: store i32 [[TMP64]], i32* [[DOTOMP_SECTIONS_IV_]], align 4 716 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 717 // CHECK2: omp.inner.for.cond: 718 // CHECK2-NEXT: [[TMP65:%.*]] = load i32, i32* [[DOTOMP_SECTIONS_IV_]], align 4 719 // CHECK2-NEXT: [[TMP66:%.*]] = load i32, i32* [[DOTOMP_SECTIONS_UB_]], align 4 720 // CHECK2-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP65]], [[TMP66]] 721 // CHECK2-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 722 // CHECK2: omp.inner.for.body: 723 // CHECK2-NEXT: [[TMP67:%.*]] = load i32, i32* [[DOTOMP_SECTIONS_IV_]], align 4 724 // CHECK2-NEXT: switch i32 [[TMP67]], label [[DOTOMP_SECTIONS_EXIT:%.*]] [ 725 // CHECK2-NEXT: i32 0, label [[DOTOMP_SECTIONS_CASE:%.*]] 726 // CHECK2-NEXT: ] 727 // CHECK2: .omp.sections.case: 728 // CHECK2-NEXT: [[TMP68:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[AGG_CAPTURED]], i32 0, i32 0 729 // CHECK2-NEXT: store i8** [[DOTTASK_RED_]], i8*** [[TMP68]], align 8 730 // CHECK2-NEXT: [[TMP69:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[AGG_CAPTURED]], i32 0, i32 1 731 // CHECK2-NEXT: store i32* [[ARGC]], i32** [[TMP69]], align 8 732 // CHECK2-NEXT: [[TMP70:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[AGG_CAPTURED]], i32 0, i32 2 733 // CHECK2-NEXT: [[TMP71:%.*]] = load i8**, i8*** [[TMP]], align 8 734 // CHECK2-NEXT: store i8** [[TMP71]], i8*** [[TMP70]], align 8 735 // CHECK2-NEXT: [[TMP72:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 736 // CHECK2-NEXT: [[TMP73:%.*]] = load i32, i32* [[TMP72]], align 4 737 // CHECK2-NEXT: [[TMP74:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP73]], 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*)*)) 738 // CHECK2-NEXT: [[TMP75:%.*]] = bitcast i8* [[TMP74]] to %struct.kmp_task_t_with_privates* 739 // CHECK2-NEXT: [[TMP76:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP75]], i32 0, i32 0 740 // CHECK2-NEXT: [[TMP77:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP76]], i32 0, i32 0 741 // CHECK2-NEXT: [[TMP78:%.*]] = load i8*, i8** [[TMP77]], align 8 742 // CHECK2-NEXT: [[TMP79:%.*]] = bitcast %struct.anon.0* [[AGG_CAPTURED]] to i8* 743 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP78]], i8* align 8 [[TMP79]], i64 24, i1 false) 744 // CHECK2-NEXT: [[TMP80:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP75]], i32 0, i32 1 745 // CHECK2-NEXT: [[TMP81:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP80]], i32 0, i32 0 746 // CHECK2-NEXT: [[TMP82:%.*]] = load i8*, i8** [[DOTTASK_RED_]], align 8 747 // CHECK2-NEXT: store i8* [[TMP82]], i8** [[TMP81]], align 8 748 // CHECK2-NEXT: [[TMP83:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 749 // CHECK2-NEXT: [[TMP84:%.*]] = load i32, i32* [[TMP83]], align 4 750 // CHECK2-NEXT: [[TMP85:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP84]], i8* [[TMP74]]) 751 // CHECK2-NEXT: br label [[DOTOMP_SECTIONS_EXIT]] 752 // CHECK2: .omp.sections.exit: 753 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 754 // CHECK2: omp.inner.for.inc: 755 // CHECK2-NEXT: [[TMP86:%.*]] = load i32, i32* [[DOTOMP_SECTIONS_IV_]], align 4 756 // CHECK2-NEXT: [[INC:%.*]] = add nsw i32 [[TMP86]], 1 757 // CHECK2-NEXT: store i32 [[INC]], i32* [[DOTOMP_SECTIONS_IV_]], align 4 758 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] 759 // CHECK2: omp.inner.for.end: 760 // CHECK2-NEXT: [[TMP87:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 761 // CHECK2-NEXT: [[TMP88:%.*]] = load i32, i32* [[TMP87]], align 4 762 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP88]]) 763 // CHECK2-NEXT: [[TMP89:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 764 // CHECK2-NEXT: [[TMP90:%.*]] = load i32, i32* [[TMP89]], align 4 765 // CHECK2-NEXT: call void @__kmpc_task_reduction_modifier_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP90]], i32 1) 766 // CHECK2-NEXT: [[TMP91:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 767 // CHECK2-NEXT: [[TMP92:%.*]] = bitcast i32* [[ARGC]] to i8* 768 // CHECK2-NEXT: store i8* [[TMP92]], i8** [[TMP91]], align 8 769 // CHECK2-NEXT: [[TMP93:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 770 // CHECK2-NEXT: store i8* [[VLA]], i8** [[TMP93]], align 8 771 // CHECK2-NEXT: [[TMP94:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 772 // CHECK2-NEXT: [[TMP95:%.*]] = inttoptr i64 [[TMP15]] to i8* 773 // CHECK2-NEXT: store i8* [[TMP95]], i8** [[TMP94]], align 8 774 // CHECK2-NEXT: [[TMP96:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 775 // CHECK2-NEXT: [[TMP97:%.*]] = load i32, i32* [[TMP96]], align 4 776 // CHECK2-NEXT: [[TMP98:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 777 // CHECK2-NEXT: [[TMP99:%.*]] = call i32 @__kmpc_reduce(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[TMP97]], i32 2, i64 24, i8* [[TMP98]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var) 778 // CHECK2-NEXT: switch i32 [[TMP99]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 779 // CHECK2-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 780 // CHECK2-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 781 // CHECK2-NEXT: ] 782 // CHECK2: .omp.reduction.case1: 783 // CHECK2-NEXT: [[TMP100:%.*]] = load i32, i32* [[TMP2]], align 4 784 // CHECK2-NEXT: [[TMP101:%.*]] = load i32, i32* [[ARGC]], align 4 785 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP100]], [[TMP101]] 786 // CHECK2-NEXT: store i32 [[ADD]], i32* [[TMP2]], align 4 787 // CHECK2-NEXT: [[TMP102:%.*]] = getelementptr i8, i8* [[ARRAYIDX1]], i64 [[TMP15]] 788 // CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[ARRAYIDX1]], [[TMP102]] 789 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE17:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 790 // CHECK2: omp.arraycpy.body: 791 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 792 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST11:%.*]] = phi i8* [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT15:%.*]], [[OMP_ARRAYCPY_BODY]] ] 793 // CHECK2-NEXT: [[TMP103:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], align 1 794 // CHECK2-NEXT: [[CONV:%.*]] = sext i8 [[TMP103]] to i32 795 // CHECK2-NEXT: [[TMP104:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 796 // CHECK2-NEXT: [[CONV12:%.*]] = sext i8 [[TMP104]] to i32 797 // CHECK2-NEXT: [[ADD13:%.*]] = add nsw i32 [[CONV]], [[CONV12]] 798 // CHECK2-NEXT: [[CONV14:%.*]] = trunc i32 [[ADD13]] to i8 799 // CHECK2-NEXT: store i8 [[CONV14]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], align 1 800 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT15]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST11]], i32 1 801 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 802 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE16:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT15]], [[TMP102]] 803 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE16]], label [[OMP_ARRAYCPY_DONE17]], label [[OMP_ARRAYCPY_BODY]] 804 // CHECK2: omp.arraycpy.done17: 805 // CHECK2-NEXT: call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP97]], [8 x i32]* @.gomp_critical_user_.reduction.var) 806 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 807 // CHECK2: .omp.reduction.case2: 808 // CHECK2-NEXT: [[TMP105:%.*]] = load i32, i32* [[ARGC]], align 4 809 // CHECK2-NEXT: [[TMP106:%.*]] = atomicrmw add i32* [[TMP2]], i32 [[TMP105]] monotonic, align 4 810 // CHECK2-NEXT: [[TMP107:%.*]] = getelementptr i8, i8* [[ARRAYIDX1]], i64 [[TMP15]] 811 // CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY18:%.*]] = icmp eq i8* [[ARRAYIDX1]], [[TMP107]] 812 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY18]], label [[OMP_ARRAYCPY_DONE31:%.*]], label [[OMP_ARRAYCPY_BODY19:%.*]] 813 // CHECK2: omp.arraycpy.body19: 814 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST20:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT29:%.*]], [[ATOMIC_EXIT:%.*]] ] 815 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST21:%.*]] = phi i8* [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT28:%.*]], [[ATOMIC_EXIT]] ] 816 // CHECK2-NEXT: [[TMP108:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST20]], align 1 817 // CHECK2-NEXT: [[CONV22:%.*]] = sext i8 [[TMP108]] to i32 818 // CHECK2-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST21]] monotonic, align 1 819 // CHECK2-NEXT: br label [[ATOMIC_CONT:%.*]] 820 // CHECK2: atomic_cont: 821 // CHECK2-NEXT: [[TMP109:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY19]] ], [ [[TMP114:%.*]], [[ATOMIC_CONT]] ] 822 // CHECK2-NEXT: store i8 [[TMP109]], i8* [[_TMP23]], align 1 823 // CHECK2-NEXT: [[TMP110:%.*]] = load i8, i8* [[_TMP23]], align 1 824 // CHECK2-NEXT: [[CONV24:%.*]] = sext i8 [[TMP110]] to i32 825 // CHECK2-NEXT: [[TMP111:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST20]], align 1 826 // CHECK2-NEXT: [[CONV25:%.*]] = sext i8 [[TMP111]] to i32 827 // CHECK2-NEXT: [[ADD26:%.*]] = add nsw i32 [[CONV24]], [[CONV25]] 828 // CHECK2-NEXT: [[CONV27:%.*]] = trunc i32 [[ADD26]] to i8 829 // CHECK2-NEXT: store i8 [[CONV27]], i8* [[ATOMIC_TEMP]], align 1 830 // CHECK2-NEXT: [[TMP112:%.*]] = load i8, i8* [[ATOMIC_TEMP]], align 1 831 // CHECK2-NEXT: [[TMP113:%.*]] = cmpxchg i8* [[OMP_ARRAYCPY_DESTELEMENTPAST21]], i8 [[TMP109]], i8 [[TMP112]] monotonic monotonic, align 1 832 // CHECK2-NEXT: [[TMP114]] = extractvalue { i8, i1 } [[TMP113]], 0 833 // CHECK2-NEXT: [[TMP115:%.*]] = extractvalue { i8, i1 } [[TMP113]], 1 834 // CHECK2-NEXT: br i1 [[TMP115]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]] 835 // CHECK2: atomic_exit: 836 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT28]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST21]], i32 1 837 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT29]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST20]], i32 1 838 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE30:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT28]], [[TMP107]] 839 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE30]], label [[OMP_ARRAYCPY_DONE31]], label [[OMP_ARRAYCPY_BODY19]] 840 // CHECK2: omp.arraycpy.done31: 841 // CHECK2-NEXT: call void @__kmpc_end_reduce(%struct.ident_t* @[[GLOB3]], i32 [[TMP97]], [8 x i32]* @.gomp_critical_user_.reduction.var) 842 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 843 // CHECK2: .omp.reduction.default: 844 // CHECK2-NEXT: [[TMP116:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 845 // CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP116]]) 846 // CHECK2-NEXT: [[TMP117:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 847 // CHECK2-NEXT: [[TMP118:%.*]] = load i32, i32* [[TMP117]], align 4 848 // CHECK2-NEXT: call void @__kmpc_barrier(%struct.ident_t* @[[GLOB4:[0-9]+]], i32 [[TMP118]]) 849 // CHECK2-NEXT: ret void 850 // 851 // 852 // CHECK2-LABEL: define {{[^@]+}}@.red_init. 853 // CHECK2-SAME: (i8* noalias [[TMP0:%.*]], i8* noalias [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { 854 // CHECK2-NEXT: entry: 855 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 856 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 857 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 858 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 859 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 860 // CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to i32* 861 // CHECK2-NEXT: store i32 0, i32* [[TMP3]], align 8 862 // CHECK2-NEXT: ret void 863 // 864 // 865 // CHECK2-LABEL: define {{[^@]+}}@.red_comb. 866 // CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 867 // CHECK2-NEXT: entry: 868 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 869 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 870 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 871 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 872 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 873 // CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to i32* 874 // CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 875 // CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to i32* 876 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP3]], align 8 877 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP5]], align 8 878 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP6]], [[TMP7]] 879 // CHECK2-NEXT: store i32 [[ADD]], i32* [[TMP3]], align 8 880 // CHECK2-NEXT: ret void 881 // 882 // 883 // CHECK2-LABEL: define {{[^@]+}}@.red_init..1 884 // CHECK2-SAME: (i8* noalias [[TMP0:%.*]], i8* noalias [[TMP1:%.*]]) #[[ATTR3]] { 885 // CHECK2-NEXT: entry: 886 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 887 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 888 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 889 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 890 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 891 // CHECK2-NEXT: [[TMP3:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8 892 // CHECK2-NEXT: [[TMP4:%.*]] = getelementptr i8, i8* [[TMP2]], i64 [[TMP3]] 893 // CHECK2-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[TMP2]], [[TMP4]] 894 // CHECK2-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 895 // CHECK2: omp.arrayinit.body: 896 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP2]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 897 // CHECK2-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 898 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 899 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]] 900 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 901 // CHECK2: omp.arrayinit.done: 902 // CHECK2-NEXT: ret void 903 // 904 // 905 // CHECK2-LABEL: define {{[^@]+}}@.red_comb..2 906 // CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 907 // CHECK2-NEXT: entry: 908 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 909 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 910 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 911 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 912 // CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8 913 // CHECK2-NEXT: [[TMP3:%.*]] = load i8*, i8** [[DOTADDR]], align 8 914 // CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 915 // CHECK2-NEXT: [[TMP5:%.*]] = getelementptr i8, i8* [[TMP3]], i64 [[TMP2]] 916 // CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP3]], [[TMP5]] 917 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 918 // CHECK2: omp.arraycpy.body: 919 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP4]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 920 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP3]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 921 // CHECK2-NEXT: [[TMP6:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 922 // CHECK2-NEXT: [[CONV:%.*]] = sext i8 [[TMP6]] to i32 923 // CHECK2-NEXT: [[TMP7:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 924 // CHECK2-NEXT: [[CONV2:%.*]] = sext i8 [[TMP7]] to i32 925 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 926 // CHECK2-NEXT: [[CONV3:%.*]] = trunc i32 [[ADD]] to i8 927 // CHECK2-NEXT: store i8 [[CONV3]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 928 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 929 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 930 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP5]] 931 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]] 932 // CHECK2: omp.arraycpy.done4: 933 // CHECK2-NEXT: ret void 934 // 935 // 936 // CHECK2-LABEL: define {{[^@]+}}@.omp_task_privates_map. 937 // CHECK2-SAME: (%struct..kmp_privates.t* noalias [[TMP0:%.*]], i8*** noalias [[TMP1:%.*]]) #[[ATTR6:[0-9]+]] { 938 // CHECK2-NEXT: entry: 939 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca %struct..kmp_privates.t*, align 8 940 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8***, align 8 941 // CHECK2-NEXT: store %struct..kmp_privates.t* [[TMP0]], %struct..kmp_privates.t** [[DOTADDR]], align 8 942 // CHECK2-NEXT: store i8*** [[TMP1]], i8**** [[DOTADDR1]], align 8 943 // CHECK2-NEXT: [[TMP2:%.*]] = load %struct..kmp_privates.t*, %struct..kmp_privates.t** [[DOTADDR]], align 8 944 // CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP2]], i32 0, i32 0 945 // CHECK2-NEXT: [[TMP4:%.*]] = load i8***, i8**** [[DOTADDR1]], align 8 946 // CHECK2-NEXT: store i8** [[TMP3]], i8*** [[TMP4]], align 8 947 // CHECK2-NEXT: ret void 948 // 949 // 950 // CHECK2-LABEL: define {{[^@]+}}@.omp_task_entry. 951 // CHECK2-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias [[TMP1:%.*]]) #[[ATTR3]] { 952 // CHECK2-NEXT: entry: 953 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4 954 // CHECK2-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8 955 // CHECK2-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8 956 // CHECK2-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8 957 // CHECK2-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8 958 // CHECK2-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon.0*, align 8 959 // CHECK2-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca i8**, align 8 960 // CHECK2-NEXT: [[TMP_I:%.*]] = alloca i8**, align 8 961 // CHECK2-NEXT: [[TMP4_I:%.*]] = alloca i8*, align 8 962 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4 963 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8 964 // CHECK2-NEXT: store i32 [[TMP0]], i32* [[DOTADDR]], align 4 965 // CHECK2-NEXT: store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8 966 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4 967 // CHECK2-NEXT: [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8 968 // CHECK2-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0 969 // CHECK2-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2 970 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0 971 // CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 972 // CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon.0* 973 // CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1 974 // CHECK2-NEXT: [[TMP10:%.*]] = bitcast %struct..kmp_privates.t* [[TMP9]] to i8* 975 // CHECK2-NEXT: [[TMP11:%.*]] = bitcast %struct.kmp_task_t_with_privates* [[TMP3]] to i8* 976 // CHECK2-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META3:![0-9]+]]) 977 // CHECK2-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META6:![0-9]+]]) 978 // CHECK2-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META8:![0-9]+]]) 979 // CHECK2-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]]) 980 // CHECK2-NEXT: store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 981 // CHECK2-NEXT: store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !12 982 // CHECK2-NEXT: store i8* [[TMP10]], i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 983 // 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 !12 984 // CHECK2-NEXT: store i8* [[TMP11]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !12 985 // CHECK2-NEXT: store %struct.anon.0* [[TMP8]], %struct.anon.0** [[__CONTEXT_ADDR_I]], align 8, !noalias !12 986 // CHECK2-NEXT: [[TMP12:%.*]] = load %struct.anon.0*, %struct.anon.0** [[__CONTEXT_ADDR_I]], align 8, !noalias !12 987 // CHECK2-NEXT: [[TMP13:%.*]] = load void (i8*, ...)*, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12 988 // CHECK2-NEXT: [[TMP14:%.*]] = load i8*, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 989 // CHECK2-NEXT: [[TMP15:%.*]] = bitcast void (i8*, ...)* [[TMP13]] to void (i8*, i8***)* 990 // CHECK2-NEXT: call void [[TMP15]](i8* [[TMP14]], i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR5:[0-9]+]] 991 // CHECK2-NEXT: [[TMP16:%.*]] = load i8**, i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias !12 992 // CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_ANON_0:%.*]], %struct.anon.0* [[TMP12]], i32 0, i32 1 993 // CHECK2-NEXT: [[TMP18:%.*]] = load i32*, i32** [[TMP17]], align 8 994 // CHECK2-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP16]], align 8 995 // CHECK2-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 996 // CHECK2-NEXT: [[TMP21:%.*]] = bitcast i32* [[TMP18]] to i8* 997 // CHECK2-NEXT: [[TMP22:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP19]], i8* [[TMP21]]) #[[ATTR5]] 998 // CHECK2-NEXT: [[CONV_I:%.*]] = bitcast i8* [[TMP22]] to i32* 999 // CHECK2-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[TMP12]], i32 0, i32 2 1000 // CHECK2-NEXT: [[TMP24:%.*]] = load i8**, i8*** [[TMP23]], align 8 1001 // CHECK2-NEXT: [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8 1002 // CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[TMP12]], i32 0, i32 1 1003 // CHECK2-NEXT: [[TMP27:%.*]] = load i32*, i32** [[TMP26]], align 8 1004 // CHECK2-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4 1005 // CHECK2-NEXT: [[TMP29:%.*]] = sext i32 [[TMP28]] to i64 1006 // CHECK2-NEXT: [[LB_ADD_LEN_I:%.*]] = add nsw i64 -1, [[TMP29]] 1007 // CHECK2-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[TMP12]], i32 0, i32 2 1008 // CHECK2-NEXT: [[TMP31:%.*]] = load i8**, i8*** [[TMP30]], align 8 1009 // CHECK2-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds i8*, i8** [[TMP31]], i64 9 1010 // CHECK2-NEXT: [[TMP32:%.*]] = load i8*, i8** [[ARRAYIDX2_I]], align 8 1011 // CHECK2-NEXT: [[ARRAYIDX3_I:%.*]] = getelementptr inbounds i8, i8* [[TMP32]], i64 [[LB_ADD_LEN_I]] 1012 // CHECK2-NEXT: [[TMP33:%.*]] = ptrtoint i8* [[ARRAYIDX3_I]] to i64 1013 // CHECK2-NEXT: [[TMP34:%.*]] = ptrtoint i8* [[TMP25]] to i64 1014 // CHECK2-NEXT: [[TMP35:%.*]] = sub i64 [[TMP33]], [[TMP34]] 1015 // CHECK2-NEXT: [[TMP36:%.*]] = sdiv exact i64 [[TMP35]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 1016 // CHECK2-NEXT: [[TMP37:%.*]] = add nuw i64 [[TMP36]], 1 1017 // CHECK2-NEXT: [[TMP38:%.*]] = mul nuw i64 [[TMP37]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 1018 // CHECK2-NEXT: store i64 [[TMP37]], i64* @{{reduction_size[.].+[.]}}, align 8 1019 // CHECK2-NEXT: [[TMP39:%.*]] = load i8*, i8** [[TMP16]], align 8 1020 // CHECK2-NEXT: [[TMP40:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP39]], i8* [[TMP25]]) #[[ATTR5]] 1021 // CHECK2-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT_ANON_0]], %struct.anon.0* [[TMP12]], i32 0, i32 2 1022 // CHECK2-NEXT: [[TMP42:%.*]] = load i8**, i8*** [[TMP41]], align 8 1023 // CHECK2-NEXT: [[TMP43:%.*]] = load i8*, i8** [[TMP42]], align 8 1024 // CHECK2-NEXT: [[TMP44:%.*]] = ptrtoint i8* [[TMP43]] to i64 1025 // CHECK2-NEXT: [[TMP45:%.*]] = ptrtoint i8* [[TMP25]] to i64 1026 // CHECK2-NEXT: [[TMP46:%.*]] = sub i64 [[TMP44]], [[TMP45]] 1027 // CHECK2-NEXT: [[TMP47:%.*]] = sdiv exact i64 [[TMP46]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 1028 // CHECK2-NEXT: [[TMP48:%.*]] = getelementptr i8, i8* [[TMP40]], i64 [[TMP47]] 1029 // CHECK2-NEXT: store i8** [[TMP4_I]], i8*** [[TMP_I]], align 8, !noalias !12 1030 // CHECK2-NEXT: store i8* [[TMP48]], i8** [[TMP4_I]], align 8, !noalias !12 1031 // CHECK2-NEXT: ret i32 0 1032 // 1033 // 1034 // CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func 1035 // CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 1036 // CHECK2-NEXT: entry: 1037 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 1038 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 1039 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 1040 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 1041 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 1042 // CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [3 x i8*]* 1043 // CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 1044 // CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [3 x i8*]* 1045 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 0 1046 // CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 1047 // CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 1048 // CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 0 1049 // CHECK2-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 1050 // CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 1051 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 1 1052 // CHECK2-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 1053 // CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1 1054 // CHECK2-NEXT: [[TMP15:%.*]] = load i8*, i8** [[TMP14]], align 8 1055 // CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2 1056 // CHECK2-NEXT: [[TMP17:%.*]] = load i8*, i8** [[TMP16]], align 8 1057 // CHECK2-NEXT: [[TMP18:%.*]] = ptrtoint i8* [[TMP17]] to i64 1058 // CHECK2-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP11]], align 4 1059 // CHECK2-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP8]], align 4 1060 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 1061 // CHECK2-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4 1062 // CHECK2-NEXT: [[TMP21:%.*]] = getelementptr i8, i8* [[TMP15]], i64 [[TMP18]] 1063 // CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP15]], [[TMP21]] 1064 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 1065 // CHECK2: omp.arraycpy.body: 1066 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP13]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 1067 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP15]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 1068 // CHECK2-NEXT: [[TMP22:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 1069 // CHECK2-NEXT: [[CONV:%.*]] = sext i8 [[TMP22]] to i32 1070 // CHECK2-NEXT: [[TMP23:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 1071 // CHECK2-NEXT: [[CONV2:%.*]] = sext i8 [[TMP23]] to i32 1072 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 1073 // CHECK2-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8 1074 // CHECK2-NEXT: store i8 [[CONV4]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 1075 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 1076 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 1077 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP21]] 1078 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]] 1079 // CHECK2: omp.arraycpy.done5: 1080 // CHECK2-NEXT: ret void 1081 // 1082