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 target parallel for reduction(task, +: argc, argv[0:10][0:argc]) 15 for (long long i = 0; i < 10; ++i) { 16 #pragma omp task in_reduction(+: argc, argv[0:10][0:argc]) 17 ; 18 } 19 } 20 21 22 23 // Init firstprivate copy of argc 24 25 // Init firstprivate copy of argv[0:10][0:argc] 26 27 // Register task reduction. 28 29 30 31 32 33 34 35 36 37 #endif 38 // CHECK1-LABEL: define {{[^@]+}}@main 39 // CHECK1-SAME: (i32 [[ARGC:%.*]], i8** [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 40 // CHECK1-NEXT: entry: 41 // CHECK1-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 42 // CHECK1-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 43 // CHECK1-NEXT: 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* nonnull align 4 dereferenceable(4) [[ARGC:%.*]], i8** [[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 [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARGC:%.*]], i8** [[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: [[DOTOMP_IV:%.*]] = alloca i64, align 8 71 // CHECK1-NEXT: [[TMP:%.*]] = alloca i64, align 8 72 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8 73 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8 74 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8 75 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 76 // CHECK1-NEXT: [[ARGC1:%.*]] = alloca i32, align 4 77 // CHECK1-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 78 // CHECK1-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 79 // CHECK1-NEXT: [[_TMP5:%.*]] = alloca i8**, align 8 80 // CHECK1-NEXT: [[_TMP6:%.*]] = alloca i8*, align 8 81 // CHECK1-NEXT: [[DOTRD_INPUT_:%.*]] = alloca [2 x %struct.kmp_taskred_input_t], align 8 82 // CHECK1-NEXT: [[DOTTASK_RED_:%.*]] = alloca i8*, align 8 83 // CHECK1-NEXT: [[I:%.*]] = alloca i64, align 8 84 // CHECK1-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8 85 // CHECK1-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8 86 // CHECK1-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1 87 // CHECK1-NEXT: [[_TMP28:%.*]] = alloca i8, align 1 88 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 89 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 90 // CHECK1-NEXT: store i32* [[ARGC]], i32** [[ARGC_ADDR]], align 8 91 // CHECK1-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 92 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARGC_ADDR]], align 8 93 // CHECK1-NEXT: store i64 0, i64* [[DOTOMP_LB]], align 8 94 // CHECK1-NEXT: store i64 9, i64* [[DOTOMP_UB]], align 8 95 // CHECK1-NEXT: store i64 1, i64* [[DOTOMP_STRIDE]], align 8 96 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 97 // CHECK1-NEXT: store i32 0, i32* [[ARGC1]], align 4 98 // CHECK1-NEXT: [[TMP1:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 99 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8*, i8** [[TMP1]], i64 0 100 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[ARRAYIDX]], align 8 101 // CHECK1-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, i8* [[TMP2]], i64 0 102 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 103 // CHECK1-NEXT: [[TMP4:%.*]] = sext i32 [[TMP3]] to i64 104 // CHECK1-NEXT: [[LB_ADD_LEN:%.*]] = add nsw i64 -1, [[TMP4]] 105 // CHECK1-NEXT: [[TMP5:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 106 // CHECK1-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds i8*, i8** [[TMP5]], i64 9 107 // CHECK1-NEXT: [[TMP6:%.*]] = load i8*, i8** [[ARRAYIDX3]], align 8 108 // CHECK1-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i8, i8* [[TMP6]], i64 [[LB_ADD_LEN]] 109 // CHECK1-NEXT: [[TMP7:%.*]] = ptrtoint i8* [[ARRAYIDX4]] to i64 110 // CHECK1-NEXT: [[TMP8:%.*]] = ptrtoint i8* [[ARRAYIDX2]] to i64 111 // CHECK1-NEXT: [[TMP9:%.*]] = sub i64 [[TMP7]], [[TMP8]] 112 // CHECK1-NEXT: [[TMP10:%.*]] = sdiv exact i64 [[TMP9]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 113 // CHECK1-NEXT: [[TMP11:%.*]] = add nuw i64 [[TMP10]], 1 114 // CHECK1-NEXT: [[TMP12:%.*]] = mul nuw i64 [[TMP11]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 115 // CHECK1-NEXT: [[TMP13:%.*]] = call i8* @llvm.stacksave() 116 // CHECK1-NEXT: store i8* [[TMP13]], i8** [[SAVED_STACK]], align 8 117 // CHECK1-NEXT: [[VLA:%.*]] = alloca i8, i64 [[TMP11]], align 16 118 // CHECK1-NEXT: store i64 [[TMP11]], i64* [[__VLA_EXPR0]], align 8 119 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP11]] 120 // CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[VLA]], [[TMP14]] 121 // CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 122 // CHECK1: omp.arrayinit.body: 123 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 124 // CHECK1-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 125 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 126 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP14]] 127 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 128 // CHECK1: omp.arrayinit.done: 129 // CHECK1-NEXT: [[TMP15:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 130 // CHECK1-NEXT: [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8 131 // CHECK1-NEXT: [[TMP17:%.*]] = ptrtoint i8* [[TMP16]] to i64 132 // CHECK1-NEXT: [[TMP18:%.*]] = ptrtoint i8* [[ARRAYIDX2]] to i64 133 // CHECK1-NEXT: [[TMP19:%.*]] = sub i64 [[TMP17]], [[TMP18]] 134 // CHECK1-NEXT: [[TMP20:%.*]] = sdiv exact i64 [[TMP19]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 135 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP20]] 136 // CHECK1-NEXT: store i8** [[_TMP6]], i8*** [[_TMP5]], align 8 137 // CHECK1-NEXT: store i8* [[TMP21]], i8** [[_TMP6]], align 8 138 // 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 0 139 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T:%.*]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 0 140 // CHECK1-NEXT: [[TMP23:%.*]] = bitcast i32* [[ARGC1]] to i8* 141 // CHECK1-NEXT: store i8* [[TMP23]], i8** [[TMP22]], align 8 142 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 1 143 // CHECK1-NEXT: [[TMP25:%.*]] = bitcast i32* [[TMP0]] to i8* 144 // CHECK1-NEXT: store i8* [[TMP25]], i8** [[TMP24]], align 8 145 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 2 146 // CHECK1-NEXT: store i64 4, i64* [[TMP26]], align 8 147 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 3 148 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init. to i8*), i8** [[TMP27]], align 8 149 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 4 150 // CHECK1-NEXT: store i8* null, i8** [[TMP28]], align 8 151 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 5 152 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb. to i8*), i8** [[TMP29]], align 8 153 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 6 154 // CHECK1-NEXT: [[TMP31:%.*]] = bitcast i32* [[TMP30]] to i8* 155 // CHECK1-NEXT: call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP31]], i8 0, i64 4, i1 false) 156 // CHECK1-NEXT: [[DOTRD_INPUT_GEP_7:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]], i64 0, i64 1 157 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 0 158 // CHECK1-NEXT: [[TMP33:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 159 // CHECK1-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i8*, i8** [[TMP33]], i64 0 160 // CHECK1-NEXT: [[TMP34:%.*]] = load i8*, i8** [[ARRAYIDX8]], align 8 161 // CHECK1-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds i8, i8* [[TMP34]], i64 0 162 // CHECK1-NEXT: [[TMP35:%.*]] = load i32, i32* [[TMP0]], align 4 163 // CHECK1-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 164 // CHECK1-NEXT: [[LB_ADD_LEN10:%.*]] = add nsw i64 -1, [[TMP36]] 165 // CHECK1-NEXT: [[TMP37:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 166 // CHECK1-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds i8*, i8** [[TMP37]], i64 9 167 // CHECK1-NEXT: [[TMP38:%.*]] = load i8*, i8** [[ARRAYIDX11]], align 8 168 // CHECK1-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds i8, i8* [[TMP38]], i64 [[LB_ADD_LEN10]] 169 // CHECK1-NEXT: store i8* [[VLA]], i8** [[TMP32]], align 8 170 // CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 1 171 // CHECK1-NEXT: store i8* [[ARRAYIDX9]], i8** [[TMP39]], align 8 172 // CHECK1-NEXT: [[TMP40:%.*]] = ptrtoint i8* [[ARRAYIDX12]] to i64 173 // CHECK1-NEXT: [[TMP41:%.*]] = ptrtoint i8* [[ARRAYIDX9]] to i64 174 // CHECK1-NEXT: [[TMP42:%.*]] = sub i64 [[TMP40]], [[TMP41]] 175 // CHECK1-NEXT: [[TMP43:%.*]] = sdiv exact i64 [[TMP42]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 176 // CHECK1-NEXT: [[TMP44:%.*]] = add nuw i64 [[TMP43]], 1 177 // CHECK1-NEXT: [[TMP45:%.*]] = mul nuw i64 [[TMP44]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 178 // CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 2 179 // CHECK1-NEXT: store i64 [[TMP45]], i64* [[TMP46]], align 8 180 // CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 3 181 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init..1 to i8*), i8** [[TMP47]], align 8 182 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 4 183 // CHECK1-NEXT: store i8* null, i8** [[TMP48]], align 8 184 // CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 5 185 // CHECK1-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb..2 to i8*), i8** [[TMP49]], align 8 186 // CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 6 187 // CHECK1-NEXT: store i32 1, i32* [[TMP50]], align 8 188 // CHECK1-NEXT: [[TMP51:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 189 // CHECK1-NEXT: [[TMP52:%.*]] = load i32, i32* [[TMP51]], align 4 190 // CHECK1-NEXT: [[TMP53:%.*]] = bitcast [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]] to i8* 191 // CHECK1-NEXT: [[TMP54:%.*]] = call i8* @__kmpc_taskred_modifier_init(%struct.ident_t* @[[GLOB1]], i32 [[TMP52]], i32 1, i32 2, i8* [[TMP53]]) 192 // CHECK1-NEXT: store i8* [[TMP54]], i8** [[DOTTASK_RED_]], align 8 193 // CHECK1-NEXT: [[TMP55:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 194 // CHECK1-NEXT: [[TMP56:%.*]] = load i32, i32* [[TMP55]], align 4 195 // CHECK1-NEXT: call void @__kmpc_for_static_init_8(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP56]], i32 34, i32* [[DOTOMP_IS_LAST]], i64* [[DOTOMP_LB]], i64* [[DOTOMP_UB]], i64* [[DOTOMP_STRIDE]], i64 1, i64 1) 196 // CHECK1-NEXT: [[TMP57:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 197 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i64 [[TMP57]], 9 198 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 199 // CHECK1: cond.true: 200 // CHECK1-NEXT: br label [[COND_END:%.*]] 201 // CHECK1: cond.false: 202 // CHECK1-NEXT: [[TMP58:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 203 // CHECK1-NEXT: br label [[COND_END]] 204 // CHECK1: cond.end: 205 // CHECK1-NEXT: [[COND:%.*]] = phi i64 [ 9, [[COND_TRUE]] ], [ [[TMP58]], [[COND_FALSE]] ] 206 // CHECK1-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8 207 // CHECK1-NEXT: [[TMP59:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8 208 // CHECK1-NEXT: store i64 [[TMP59]], i64* [[DOTOMP_IV]], align 8 209 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 210 // CHECK1: omp.inner.for.cond: 211 // CHECK1-NEXT: [[TMP60:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 212 // CHECK1-NEXT: [[TMP61:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 213 // CHECK1-NEXT: [[CMP13:%.*]] = icmp sle i64 [[TMP60]], [[TMP61]] 214 // CHECK1-NEXT: br i1 [[CMP13]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]] 215 // CHECK1: omp.inner.for.cond.cleanup: 216 // CHECK1-NEXT: br label [[OMP_INNER_FOR_END:%.*]] 217 // CHECK1: omp.inner.for.body: 218 // CHECK1-NEXT: [[TMP62:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 219 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i64 [[TMP62]], 1 220 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL]] 221 // CHECK1-NEXT: store i64 [[ADD]], i64* [[I]], align 8 222 // CHECK1-NEXT: [[TMP63:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 0 223 // CHECK1-NEXT: store i8** [[DOTTASK_RED_]], i8*** [[TMP63]], align 8 224 // CHECK1-NEXT: [[TMP64:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 1 225 // CHECK1-NEXT: store i32* [[ARGC1]], i32** [[TMP64]], align 8 226 // CHECK1-NEXT: [[TMP65:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 2 227 // CHECK1-NEXT: [[TMP66:%.*]] = load i8**, i8*** [[_TMP5]], align 8 228 // CHECK1-NEXT: store i8** [[TMP66]], i8*** [[TMP65]], align 8 229 // CHECK1-NEXT: [[TMP67:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 230 // CHECK1-NEXT: [[TMP68:%.*]] = load i32, i32* [[TMP67]], align 4 231 // CHECK1-NEXT: [[TMP69:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP68]], 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*)*)) 232 // CHECK1-NEXT: [[TMP70:%.*]] = bitcast i8* [[TMP69]] to %struct.kmp_task_t_with_privates* 233 // CHECK1-NEXT: [[TMP71:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP70]], i32 0, i32 0 234 // CHECK1-NEXT: [[TMP72:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP71]], i32 0, i32 0 235 // CHECK1-NEXT: [[TMP73:%.*]] = load i8*, i8** [[TMP72]], align 8 236 // CHECK1-NEXT: [[TMP74:%.*]] = bitcast %struct.anon* [[AGG_CAPTURED]] to i8* 237 // CHECK1-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP73]], i8* align 8 [[TMP74]], i64 24, i1 false) 238 // CHECK1-NEXT: [[TMP75:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP70]], i32 0, i32 1 239 // CHECK1-NEXT: [[TMP76:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP75]], i32 0, i32 0 240 // CHECK1-NEXT: [[TMP77:%.*]] = load i8*, i8** [[DOTTASK_RED_]], align 8 241 // CHECK1-NEXT: store i8* [[TMP77]], i8** [[TMP76]], align 8 242 // CHECK1-NEXT: [[TMP78:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 243 // CHECK1-NEXT: [[TMP79:%.*]] = load i32, i32* [[TMP78]], align 4 244 // CHECK1-NEXT: [[TMP80:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP79]], i8* [[TMP69]]) 245 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 246 // CHECK1: omp.body.continue: 247 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 248 // CHECK1: omp.inner.for.inc: 249 // CHECK1-NEXT: [[TMP81:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 250 // CHECK1-NEXT: [[ADD14:%.*]] = add nsw i64 [[TMP81]], 1 251 // CHECK1-NEXT: store i64 [[ADD14]], i64* [[DOTOMP_IV]], align 8 252 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 253 // CHECK1: omp.inner.for.end: 254 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 255 // CHECK1: omp.loop.exit: 256 // CHECK1-NEXT: [[TMP82:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 257 // CHECK1-NEXT: [[TMP83:%.*]] = load i32, i32* [[TMP82]], align 4 258 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP83]]) 259 // CHECK1-NEXT: [[TMP84:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 260 // CHECK1-NEXT: [[TMP85:%.*]] = load i32, i32* [[TMP84]], align 4 261 // CHECK1-NEXT: call void @__kmpc_task_reduction_modifier_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP85]], i32 1) 262 // CHECK1-NEXT: [[TMP86:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 263 // CHECK1-NEXT: [[TMP87:%.*]] = bitcast i32* [[ARGC1]] to i8* 264 // CHECK1-NEXT: store i8* [[TMP87]], i8** [[TMP86]], align 8 265 // CHECK1-NEXT: [[TMP88:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 266 // CHECK1-NEXT: store i8* [[VLA]], i8** [[TMP88]], align 8 267 // CHECK1-NEXT: [[TMP89:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 268 // CHECK1-NEXT: [[TMP90:%.*]] = inttoptr i64 [[TMP11]] to i8* 269 // CHECK1-NEXT: store i8* [[TMP90]], i8** [[TMP89]], align 8 270 // CHECK1-NEXT: [[TMP91:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 271 // CHECK1-NEXT: [[TMP92:%.*]] = load i32, i32* [[TMP91]], align 4 272 // CHECK1-NEXT: [[TMP93:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 273 // CHECK1-NEXT: [[TMP94:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[TMP92]], i32 2, i64 24, i8* [[TMP93]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var) 274 // CHECK1-NEXT: switch i32 [[TMP94]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 275 // CHECK1-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 276 // CHECK1-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 277 // CHECK1-NEXT: ] 278 // CHECK1: .omp.reduction.case1: 279 // CHECK1-NEXT: [[TMP95:%.*]] = load i32, i32* [[TMP0]], align 4 280 // CHECK1-NEXT: [[TMP96:%.*]] = load i32, i32* [[ARGC1]], align 4 281 // CHECK1-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP95]], [[TMP96]] 282 // CHECK1-NEXT: store i32 [[ADD15]], i32* [[TMP0]], align 4 283 // CHECK1-NEXT: [[TMP97:%.*]] = getelementptr i8, i8* [[ARRAYIDX2]], i64 [[TMP11]] 284 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[ARRAYIDX2]], [[TMP97]] 285 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE22:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 286 // CHECK1: omp.arraycpy.body: 287 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 288 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST16:%.*]] = phi i8* [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT20:%.*]], [[OMP_ARRAYCPY_BODY]] ] 289 // CHECK1-NEXT: [[TMP98:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1 290 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP98]] to i32 291 // CHECK1-NEXT: [[TMP99:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 292 // CHECK1-NEXT: [[CONV17:%.*]] = sext i8 [[TMP99]] to i32 293 // CHECK1-NEXT: [[ADD18:%.*]] = add nsw i32 [[CONV]], [[CONV17]] 294 // CHECK1-NEXT: [[CONV19:%.*]] = trunc i32 [[ADD18]] to i8 295 // CHECK1-NEXT: store i8 [[CONV19]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1 296 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT20]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST16]], i32 1 297 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 298 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE21:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT20]], [[TMP97]] 299 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_DONE22]], label [[OMP_ARRAYCPY_BODY]] 300 // CHECK1: omp.arraycpy.done22: 301 // CHECK1-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP92]], [8 x i32]* @.gomp_critical_user_.reduction.var) 302 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 303 // CHECK1: .omp.reduction.case2: 304 // CHECK1-NEXT: [[TMP100:%.*]] = load i32, i32* [[ARGC1]], align 4 305 // CHECK1-NEXT: [[TMP101:%.*]] = atomicrmw add i32* [[TMP0]], i32 [[TMP100]] monotonic, align 4 306 // CHECK1-NEXT: [[TMP102:%.*]] = getelementptr i8, i8* [[ARRAYIDX2]], i64 [[TMP11]] 307 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY23:%.*]] = icmp eq i8* [[ARRAYIDX2]], [[TMP102]] 308 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY23]], label [[OMP_ARRAYCPY_DONE36:%.*]], label [[OMP_ARRAYCPY_BODY24:%.*]] 309 // CHECK1: omp.arraycpy.body24: 310 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST25:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT34:%.*]], [[ATOMIC_EXIT:%.*]] ] 311 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST26:%.*]] = phi i8* [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT33:%.*]], [[ATOMIC_EXIT]] ] 312 // CHECK1-NEXT: [[TMP103:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1 313 // CHECK1-NEXT: [[CONV27:%.*]] = sext i8 [[TMP103]] to i32 314 // CHECK1-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST26]] monotonic, align 1 315 // CHECK1-NEXT: br label [[ATOMIC_CONT:%.*]] 316 // CHECK1: atomic_cont: 317 // CHECK1-NEXT: [[TMP104:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY24]] ], [ [[TMP109:%.*]], [[ATOMIC_CONT]] ] 318 // CHECK1-NEXT: store i8 [[TMP104]], i8* [[_TMP28]], align 1 319 // CHECK1-NEXT: [[TMP105:%.*]] = load i8, i8* [[_TMP28]], align 1 320 // CHECK1-NEXT: [[CONV29:%.*]] = sext i8 [[TMP105]] to i32 321 // CHECK1-NEXT: [[TMP106:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1 322 // CHECK1-NEXT: [[CONV30:%.*]] = sext i8 [[TMP106]] to i32 323 // CHECK1-NEXT: [[ADD31:%.*]] = add nsw i32 [[CONV29]], [[CONV30]] 324 // CHECK1-NEXT: [[CONV32:%.*]] = trunc i32 [[ADD31]] to i8 325 // CHECK1-NEXT: store i8 [[CONV32]], i8* [[ATOMIC_TEMP]], align 1 326 // CHECK1-NEXT: [[TMP107:%.*]] = load i8, i8* [[ATOMIC_TEMP]], align 1 327 // CHECK1-NEXT: [[TMP108:%.*]] = cmpxchg i8* [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i8 [[TMP104]], i8 [[TMP107]] monotonic monotonic, align 1 328 // CHECK1-NEXT: [[TMP109]] = extractvalue { i8, i1 } [[TMP108]], 0 329 // CHECK1-NEXT: [[TMP110:%.*]] = extractvalue { i8, i1 } [[TMP108]], 1 330 // CHECK1-NEXT: br i1 [[TMP110]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]] 331 // CHECK1: atomic_exit: 332 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT33]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i32 1 333 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT34]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST25]], i32 1 334 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE35:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT33]], [[TMP102]] 335 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE35]], label [[OMP_ARRAYCPY_DONE36]], label [[OMP_ARRAYCPY_BODY24]] 336 // CHECK1: omp.arraycpy.done36: 337 // CHECK1-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 338 // CHECK1: .omp.reduction.default: 339 // CHECK1-NEXT: [[TMP111:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 340 // CHECK1-NEXT: call void @llvm.stackrestore(i8* [[TMP111]]) 341 // CHECK1-NEXT: ret void 342 // 343 // 344 // CHECK1-LABEL: define {{[^@]+}}@.red_init. 345 // CHECK1-SAME: (i8* noalias [[TMP0:%.*]], i8* noalias [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { 346 // CHECK1-NEXT: entry: 347 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 348 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 349 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 350 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 351 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 352 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to i32* 353 // CHECK1-NEXT: store i32 0, i32* [[TMP3]], align 8 354 // CHECK1-NEXT: ret void 355 // 356 // 357 // CHECK1-LABEL: define {{[^@]+}}@.red_comb. 358 // CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 359 // CHECK1-NEXT: entry: 360 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 361 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 362 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 363 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 364 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 365 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to i32* 366 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 367 // CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to i32* 368 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP3]], align 8 369 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP5]], align 8 370 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP6]], [[TMP7]] 371 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP3]], align 8 372 // CHECK1-NEXT: ret void 373 // 374 // 375 // CHECK1-LABEL: define {{[^@]+}}@.red_init..1 376 // CHECK1-SAME: (i8* noalias [[TMP0:%.*]], i8* noalias [[TMP1:%.*]]) #[[ATTR3]] { 377 // CHECK1-NEXT: entry: 378 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 379 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 380 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 381 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 382 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 383 // CHECK1-NEXT: [[TMP3:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8 384 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr i8, i8* [[TMP2]], i64 [[TMP3]] 385 // CHECK1-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[TMP2]], [[TMP4]] 386 // CHECK1-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 387 // CHECK1: omp.arrayinit.body: 388 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP2]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 389 // CHECK1-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 390 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 391 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]] 392 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 393 // CHECK1: omp.arrayinit.done: 394 // CHECK1-NEXT: ret void 395 // 396 // 397 // CHECK1-LABEL: define {{[^@]+}}@.red_comb..2 398 // CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 399 // CHECK1-NEXT: entry: 400 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 401 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 402 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 403 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 404 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8 405 // CHECK1-NEXT: [[TMP3:%.*]] = load i8*, i8** [[DOTADDR]], align 8 406 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 407 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr i8, i8* [[TMP3]], i64 [[TMP2]] 408 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP3]], [[TMP5]] 409 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 410 // CHECK1: omp.arraycpy.body: 411 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP4]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 412 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP3]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 413 // CHECK1-NEXT: [[TMP6:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 414 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP6]] to i32 415 // CHECK1-NEXT: [[TMP7:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 416 // CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP7]] to i32 417 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 418 // CHECK1-NEXT: [[CONV3:%.*]] = trunc i32 [[ADD]] to i8 419 // CHECK1-NEXT: store i8 [[CONV3]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 420 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 421 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 422 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP5]] 423 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]] 424 // CHECK1: omp.arraycpy.done4: 425 // CHECK1-NEXT: ret void 426 // 427 // 428 // CHECK1-LABEL: define {{[^@]+}}@.omp_task_privates_map. 429 // CHECK1-SAME: (%struct..kmp_privates.t* noalias [[TMP0:%.*]], i8*** noalias [[TMP1:%.*]]) #[[ATTR6:[0-9]+]] { 430 // CHECK1-NEXT: entry: 431 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca %struct..kmp_privates.t*, align 8 432 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8***, align 8 433 // CHECK1-NEXT: store %struct..kmp_privates.t* [[TMP0]], %struct..kmp_privates.t** [[DOTADDR]], align 8 434 // CHECK1-NEXT: store i8*** [[TMP1]], i8**** [[DOTADDR1]], align 8 435 // CHECK1-NEXT: [[TMP2:%.*]] = load %struct..kmp_privates.t*, %struct..kmp_privates.t** [[DOTADDR]], align 8 436 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP2]], i32 0, i32 0 437 // CHECK1-NEXT: [[TMP4:%.*]] = load i8***, i8**** [[DOTADDR1]], align 8 438 // CHECK1-NEXT: store i8** [[TMP3]], i8*** [[TMP4]], align 8 439 // CHECK1-NEXT: ret void 440 // 441 // 442 // CHECK1-LABEL: define {{[^@]+}}@.omp_task_entry. 443 // CHECK1-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias [[TMP1:%.*]]) #[[ATTR3]] { 444 // CHECK1-NEXT: entry: 445 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4 446 // CHECK1-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8 447 // CHECK1-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8 448 // CHECK1-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8 449 // CHECK1-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8 450 // CHECK1-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon*, align 8 451 // CHECK1-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca i8**, align 8 452 // CHECK1-NEXT: [[TMP_I:%.*]] = alloca i8**, align 8 453 // CHECK1-NEXT: [[TMP4_I:%.*]] = alloca i8*, align 8 454 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4 455 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8 456 // CHECK1-NEXT: store i32 [[TMP0]], i32* [[DOTADDR]], align 4 457 // CHECK1-NEXT: store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8 458 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4 459 // CHECK1-NEXT: [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8 460 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0 461 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2 462 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0 463 // CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 464 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon* 465 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1 466 // CHECK1-NEXT: [[TMP10:%.*]] = bitcast %struct..kmp_privates.t* [[TMP9]] to i8* 467 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast %struct.kmp_task_t_with_privates* [[TMP3]] to i8* 468 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META3:![0-9]+]]) 469 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META6:![0-9]+]]) 470 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META8:![0-9]+]]) 471 // CHECK1-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]]) 472 // CHECK1-NEXT: store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 473 // CHECK1-NEXT: store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !12 474 // CHECK1-NEXT: store i8* [[TMP10]], i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 475 // 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 !12 476 // CHECK1-NEXT: store i8* [[TMP11]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !12 477 // CHECK1-NEXT: store %struct.anon* [[TMP8]], %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !12 478 // CHECK1-NEXT: [[TMP12:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !12 479 // CHECK1-NEXT: [[TMP13:%.*]] = load void (i8*, ...)*, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12 480 // CHECK1-NEXT: [[TMP14:%.*]] = load i8*, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 481 // CHECK1-NEXT: [[TMP15:%.*]] = bitcast void (i8*, ...)* [[TMP13]] to void (i8*, i8***)* 482 // CHECK1-NEXT: call void [[TMP15]](i8* [[TMP14]], i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR5]] 483 // CHECK1-NEXT: [[TMP16:%.*]] = load i8**, i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias !12 484 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP12]], i32 0, i32 1 485 // CHECK1-NEXT: [[TMP18:%.*]] = load i32*, i32** [[TMP17]], align 8 486 // CHECK1-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP16]], align 8 487 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 488 // CHECK1-NEXT: [[TMP21:%.*]] = bitcast i32* [[TMP18]] to i8* 489 // CHECK1-NEXT: [[TMP22:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP19]], i8* [[TMP21]]) #[[ATTR5]] 490 // CHECK1-NEXT: [[CONV_I:%.*]] = bitcast i8* [[TMP22]] to i32* 491 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2 492 // CHECK1-NEXT: [[TMP24:%.*]] = load i8**, i8*** [[TMP23]], align 8 493 // CHECK1-NEXT: [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8 494 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 1 495 // CHECK1-NEXT: [[TMP27:%.*]] = load i32*, i32** [[TMP26]], align 8 496 // CHECK1-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4 497 // CHECK1-NEXT: [[TMP29:%.*]] = sext i32 [[TMP28]] to i64 498 // CHECK1-NEXT: [[LB_ADD_LEN_I:%.*]] = add nsw i64 -1, [[TMP29]] 499 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2 500 // CHECK1-NEXT: [[TMP31:%.*]] = load i8**, i8*** [[TMP30]], align 8 501 // CHECK1-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds i8*, i8** [[TMP31]], i64 9 502 // CHECK1-NEXT: [[TMP32:%.*]] = load i8*, i8** [[ARRAYIDX2_I]], align 8 503 // CHECK1-NEXT: [[ARRAYIDX3_I:%.*]] = getelementptr inbounds i8, i8* [[TMP32]], i64 [[LB_ADD_LEN_I]] 504 // CHECK1-NEXT: [[TMP33:%.*]] = ptrtoint i8* [[ARRAYIDX3_I]] to i64 505 // CHECK1-NEXT: [[TMP34:%.*]] = ptrtoint i8* [[TMP25]] to i64 506 // CHECK1-NEXT: [[TMP35:%.*]] = sub i64 [[TMP33]], [[TMP34]] 507 // CHECK1-NEXT: [[TMP36:%.*]] = sdiv exact i64 [[TMP35]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 508 // CHECK1-NEXT: [[TMP37:%.*]] = add nuw i64 [[TMP36]], 1 509 // CHECK1-NEXT: [[TMP38:%.*]] = mul nuw i64 [[TMP37]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 510 // CHECK1-NEXT: store i64 [[TMP37]], i64* @{{reduction_size[.].+[.]}}, align 8 511 // CHECK1-NEXT: [[TMP39:%.*]] = load i8*, i8** [[TMP16]], align 8 512 // CHECK1-NEXT: [[TMP40:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP39]], i8* [[TMP25]]) #[[ATTR5]] 513 // CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2 514 // CHECK1-NEXT: [[TMP42:%.*]] = load i8**, i8*** [[TMP41]], align 8 515 // CHECK1-NEXT: [[TMP43:%.*]] = load i8*, i8** [[TMP42]], align 8 516 // CHECK1-NEXT: [[TMP44:%.*]] = ptrtoint i8* [[TMP43]] to i64 517 // CHECK1-NEXT: [[TMP45:%.*]] = ptrtoint i8* [[TMP25]] to i64 518 // CHECK1-NEXT: [[TMP46:%.*]] = sub i64 [[TMP44]], [[TMP45]] 519 // CHECK1-NEXT: [[TMP47:%.*]] = sdiv exact i64 [[TMP46]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 520 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr i8, i8* [[TMP40]], i64 [[TMP47]] 521 // CHECK1-NEXT: store i8** [[TMP4_I]], i8*** [[TMP_I]], align 8, !noalias !12 522 // CHECK1-NEXT: store i8* [[TMP48]], i8** [[TMP4_I]], align 8, !noalias !12 523 // CHECK1-NEXT: ret i32 0 524 // 525 // 526 // CHECK1-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func 527 // CHECK1-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 528 // CHECK1-NEXT: entry: 529 // CHECK1-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 530 // CHECK1-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 531 // CHECK1-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 532 // CHECK1-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 533 // CHECK1-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 534 // CHECK1-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [3 x i8*]* 535 // CHECK1-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 536 // CHECK1-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [3 x i8*]* 537 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 0 538 // CHECK1-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 539 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 540 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 0 541 // CHECK1-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 542 // CHECK1-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 543 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 1 544 // CHECK1-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 545 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1 546 // CHECK1-NEXT: [[TMP15:%.*]] = load i8*, i8** [[TMP14]], align 8 547 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2 548 // CHECK1-NEXT: [[TMP17:%.*]] = load i8*, i8** [[TMP16]], align 8 549 // CHECK1-NEXT: [[TMP18:%.*]] = ptrtoint i8* [[TMP17]] to i64 550 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP11]], align 4 551 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP8]], align 4 552 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 553 // CHECK1-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4 554 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr i8, i8* [[TMP15]], i64 [[TMP18]] 555 // CHECK1-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP15]], [[TMP21]] 556 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 557 // CHECK1: omp.arraycpy.body: 558 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP13]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 559 // CHECK1-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP15]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 560 // CHECK1-NEXT: [[TMP22:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 561 // CHECK1-NEXT: [[CONV:%.*]] = sext i8 [[TMP22]] to i32 562 // CHECK1-NEXT: [[TMP23:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 563 // CHECK1-NEXT: [[CONV2:%.*]] = sext i8 [[TMP23]] to i32 564 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 565 // CHECK1-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8 566 // CHECK1-NEXT: store i8 [[CONV4]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 567 // CHECK1-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 568 // CHECK1-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 569 // CHECK1-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP21]] 570 // CHECK1-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]] 571 // CHECK1: omp.arraycpy.done5: 572 // CHECK1-NEXT: ret void 573 // 574 // 575 // CHECK2-LABEL: define {{[^@]+}}@main 576 // CHECK2-SAME: (i32 [[ARGC:%.*]], i8** [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 577 // CHECK2-NEXT: entry: 578 // CHECK2-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 579 // CHECK2-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 580 // CHECK2-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 581 // CHECK2-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 582 // CHECK2-NEXT: [[TMP0:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 583 // CHECK2-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l14(i32* [[ARGC_ADDR]], i8** [[TMP0]]) #[[ATTR5:[0-9]+]] 584 // CHECK2-NEXT: ret i32 0 585 // 586 // 587 // CHECK2-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l14 588 // CHECK2-SAME: (i32* nonnull align 4 dereferenceable(4) [[ARGC:%.*]], i8** [[ARGV:%.*]]) #[[ATTR1:[0-9]+]] { 589 // CHECK2-NEXT: entry: 590 // CHECK2-NEXT: [[ARGC_ADDR:%.*]] = alloca i32*, align 8 591 // CHECK2-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 592 // CHECK2-NEXT: store i32* [[ARGC]], i32** [[ARGC_ADDR]], align 8 593 // CHECK2-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 594 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARGC_ADDR]], align 8 595 // CHECK2-NEXT: [[TMP1:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 596 // CHECK2-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]]) 597 // CHECK2-NEXT: ret void 598 // 599 // 600 // CHECK2-LABEL: define {{[^@]+}}@.omp_outlined. 601 // CHECK2-SAME: (i32* noalias [[DOTGLOBAL_TID_:%.*]], i32* noalias [[DOTBOUND_TID_:%.*]], i32* nonnull align 4 dereferenceable(4) [[ARGC:%.*]], i8** [[ARGV:%.*]]) #[[ATTR1]] { 602 // CHECK2-NEXT: entry: 603 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 604 // CHECK2-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 605 // CHECK2-NEXT: [[ARGC_ADDR:%.*]] = alloca i32*, align 8 606 // CHECK2-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 607 // CHECK2-NEXT: [[DOTOMP_IV:%.*]] = alloca i64, align 8 608 // CHECK2-NEXT: [[TMP:%.*]] = alloca i64, align 8 609 // CHECK2-NEXT: [[DOTOMP_LB:%.*]] = alloca i64, align 8 610 // CHECK2-NEXT: [[DOTOMP_UB:%.*]] = alloca i64, align 8 611 // CHECK2-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i64, align 8 612 // CHECK2-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 613 // CHECK2-NEXT: [[ARGC1:%.*]] = alloca i32, align 4 614 // CHECK2-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 615 // CHECK2-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 616 // CHECK2-NEXT: [[_TMP5:%.*]] = alloca i8**, align 8 617 // CHECK2-NEXT: [[_TMP6:%.*]] = alloca i8*, align 8 618 // CHECK2-NEXT: [[DOTRD_INPUT_:%.*]] = alloca [2 x %struct.kmp_taskred_input_t], align 8 619 // CHECK2-NEXT: [[DOTTASK_RED_:%.*]] = alloca i8*, align 8 620 // CHECK2-NEXT: [[I:%.*]] = alloca i64, align 8 621 // CHECK2-NEXT: [[AGG_CAPTURED:%.*]] = alloca [[STRUCT_ANON:%.*]], align 8 622 // CHECK2-NEXT: [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x i8*], align 8 623 // CHECK2-NEXT: [[ATOMIC_TEMP:%.*]] = alloca i8, align 1 624 // CHECK2-NEXT: [[_TMP28:%.*]] = alloca i8, align 1 625 // CHECK2-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 626 // CHECK2-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 627 // CHECK2-NEXT: store i32* [[ARGC]], i32** [[ARGC_ADDR]], align 8 628 // CHECK2-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 629 // CHECK2-NEXT: [[TMP0:%.*]] = load i32*, i32** [[ARGC_ADDR]], align 8 630 // CHECK2-NEXT: store i64 0, i64* [[DOTOMP_LB]], align 8 631 // CHECK2-NEXT: store i64 9, i64* [[DOTOMP_UB]], align 8 632 // CHECK2-NEXT: store i64 1, i64* [[DOTOMP_STRIDE]], align 8 633 // CHECK2-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 634 // CHECK2-NEXT: store i32 0, i32* [[ARGC1]], align 4 635 // CHECK2-NEXT: [[TMP1:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 636 // CHECK2-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i8*, i8** [[TMP1]], i64 0 637 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[ARRAYIDX]], align 8 638 // CHECK2-NEXT: [[ARRAYIDX2:%.*]] = getelementptr inbounds i8, i8* [[TMP2]], i64 0 639 // CHECK2-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 640 // CHECK2-NEXT: [[TMP4:%.*]] = sext i32 [[TMP3]] to i64 641 // CHECK2-NEXT: [[LB_ADD_LEN:%.*]] = add nsw i64 -1, [[TMP4]] 642 // CHECK2-NEXT: [[TMP5:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 643 // CHECK2-NEXT: [[ARRAYIDX3:%.*]] = getelementptr inbounds i8*, i8** [[TMP5]], i64 9 644 // CHECK2-NEXT: [[TMP6:%.*]] = load i8*, i8** [[ARRAYIDX3]], align 8 645 // CHECK2-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds i8, i8* [[TMP6]], i64 [[LB_ADD_LEN]] 646 // CHECK2-NEXT: [[TMP7:%.*]] = ptrtoint i8* [[ARRAYIDX4]] to i64 647 // CHECK2-NEXT: [[TMP8:%.*]] = ptrtoint i8* [[ARRAYIDX2]] to i64 648 // CHECK2-NEXT: [[TMP9:%.*]] = sub i64 [[TMP7]], [[TMP8]] 649 // CHECK2-NEXT: [[TMP10:%.*]] = sdiv exact i64 [[TMP9]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 650 // CHECK2-NEXT: [[TMP11:%.*]] = add nuw i64 [[TMP10]], 1 651 // CHECK2-NEXT: [[TMP12:%.*]] = mul nuw i64 [[TMP11]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 652 // CHECK2-NEXT: [[TMP13:%.*]] = call i8* @llvm.stacksave() 653 // CHECK2-NEXT: store i8* [[TMP13]], i8** [[SAVED_STACK]], align 8 654 // CHECK2-NEXT: [[VLA:%.*]] = alloca i8, i64 [[TMP11]], align 16 655 // CHECK2-NEXT: store i64 [[TMP11]], i64* [[__VLA_EXPR0]], align 8 656 // CHECK2-NEXT: [[TMP14:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP11]] 657 // CHECK2-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[VLA]], [[TMP14]] 658 // CHECK2-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 659 // CHECK2: omp.arrayinit.body: 660 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 661 // CHECK2-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 662 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 663 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP14]] 664 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 665 // CHECK2: omp.arrayinit.done: 666 // CHECK2-NEXT: [[TMP15:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 667 // CHECK2-NEXT: [[TMP16:%.*]] = load i8*, i8** [[TMP15]], align 8 668 // CHECK2-NEXT: [[TMP17:%.*]] = ptrtoint i8* [[TMP16]] to i64 669 // CHECK2-NEXT: [[TMP18:%.*]] = ptrtoint i8* [[ARRAYIDX2]] to i64 670 // CHECK2-NEXT: [[TMP19:%.*]] = sub i64 [[TMP17]], [[TMP18]] 671 // CHECK2-NEXT: [[TMP20:%.*]] = sdiv exact i64 [[TMP19]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 672 // CHECK2-NEXT: [[TMP21:%.*]] = getelementptr i8, i8* [[VLA]], i64 [[TMP20]] 673 // CHECK2-NEXT: store i8** [[_TMP6]], i8*** [[_TMP5]], align 8 674 // CHECK2-NEXT: store i8* [[TMP21]], i8** [[_TMP6]], align 8 675 // 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 0 676 // CHECK2-NEXT: [[TMP22:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T:%.*]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 0 677 // CHECK2-NEXT: [[TMP23:%.*]] = bitcast i32* [[ARGC1]] to i8* 678 // CHECK2-NEXT: store i8* [[TMP23]], i8** [[TMP22]], align 8 679 // CHECK2-NEXT: [[TMP24:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 1 680 // CHECK2-NEXT: [[TMP25:%.*]] = bitcast i32* [[TMP0]] to i8* 681 // CHECK2-NEXT: store i8* [[TMP25]], i8** [[TMP24]], align 8 682 // CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 2 683 // CHECK2-NEXT: store i64 4, i64* [[TMP26]], align 8 684 // CHECK2-NEXT: [[TMP27:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 3 685 // CHECK2-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init. to i8*), i8** [[TMP27]], align 8 686 // CHECK2-NEXT: [[TMP28:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 4 687 // CHECK2-NEXT: store i8* null, i8** [[TMP28]], align 8 688 // CHECK2-NEXT: [[TMP29:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 5 689 // CHECK2-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb. to i8*), i8** [[TMP29]], align 8 690 // CHECK2-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_]], i32 0, i32 6 691 // CHECK2-NEXT: [[TMP31:%.*]] = bitcast i32* [[TMP30]] to i8* 692 // CHECK2-NEXT: call void @llvm.memset.p0i8.i64(i8* align 8 [[TMP31]], i8 0, i64 4, i1 false) 693 // CHECK2-NEXT: [[DOTRD_INPUT_GEP_7:%.*]] = getelementptr inbounds [2 x %struct.kmp_taskred_input_t], [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]], i64 0, i64 1 694 // CHECK2-NEXT: [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 0 695 // CHECK2-NEXT: [[TMP33:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 696 // CHECK2-NEXT: [[ARRAYIDX8:%.*]] = getelementptr inbounds i8*, i8** [[TMP33]], i64 0 697 // CHECK2-NEXT: [[TMP34:%.*]] = load i8*, i8** [[ARRAYIDX8]], align 8 698 // CHECK2-NEXT: [[ARRAYIDX9:%.*]] = getelementptr inbounds i8, i8* [[TMP34]], i64 0 699 // CHECK2-NEXT: [[TMP35:%.*]] = load i32, i32* [[TMP0]], align 4 700 // CHECK2-NEXT: [[TMP36:%.*]] = sext i32 [[TMP35]] to i64 701 // CHECK2-NEXT: [[LB_ADD_LEN10:%.*]] = add nsw i64 -1, [[TMP36]] 702 // CHECK2-NEXT: [[TMP37:%.*]] = load i8**, i8*** [[ARGV_ADDR]], align 8 703 // CHECK2-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds i8*, i8** [[TMP37]], i64 9 704 // CHECK2-NEXT: [[TMP38:%.*]] = load i8*, i8** [[ARRAYIDX11]], align 8 705 // CHECK2-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds i8, i8* [[TMP38]], i64 [[LB_ADD_LEN10]] 706 // CHECK2-NEXT: store i8* [[VLA]], i8** [[TMP32]], align 8 707 // CHECK2-NEXT: [[TMP39:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 1 708 // CHECK2-NEXT: store i8* [[ARRAYIDX9]], i8** [[TMP39]], align 8 709 // CHECK2-NEXT: [[TMP40:%.*]] = ptrtoint i8* [[ARRAYIDX12]] to i64 710 // CHECK2-NEXT: [[TMP41:%.*]] = ptrtoint i8* [[ARRAYIDX9]] to i64 711 // CHECK2-NEXT: [[TMP42:%.*]] = sub i64 [[TMP40]], [[TMP41]] 712 // CHECK2-NEXT: [[TMP43:%.*]] = sdiv exact i64 [[TMP42]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 713 // CHECK2-NEXT: [[TMP44:%.*]] = add nuw i64 [[TMP43]], 1 714 // CHECK2-NEXT: [[TMP45:%.*]] = mul nuw i64 [[TMP44]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 715 // CHECK2-NEXT: [[TMP46:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 2 716 // CHECK2-NEXT: store i64 [[TMP45]], i64* [[TMP46]], align 8 717 // CHECK2-NEXT: [[TMP47:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 3 718 // CHECK2-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_init..1 to i8*), i8** [[TMP47]], align 8 719 // CHECK2-NEXT: [[TMP48:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 4 720 // CHECK2-NEXT: store i8* null, i8** [[TMP48]], align 8 721 // CHECK2-NEXT: [[TMP49:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 5 722 // CHECK2-NEXT: store i8* bitcast (void (i8*, i8*)* @.red_comb..2 to i8*), i8** [[TMP49]], align 8 723 // CHECK2-NEXT: [[TMP50:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASKRED_INPUT_T]], %struct.kmp_taskred_input_t* [[DOTRD_INPUT_GEP_7]], i32 0, i32 6 724 // CHECK2-NEXT: store i32 1, i32* [[TMP50]], align 8 725 // CHECK2-NEXT: [[TMP51:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 726 // CHECK2-NEXT: [[TMP52:%.*]] = load i32, i32* [[TMP51]], align 4 727 // CHECK2-NEXT: [[TMP53:%.*]] = bitcast [2 x %struct.kmp_taskred_input_t]* [[DOTRD_INPUT_]] to i8* 728 // CHECK2-NEXT: [[TMP54:%.*]] = call i8* @__kmpc_taskred_modifier_init(%struct.ident_t* @[[GLOB1]], i32 [[TMP52]], i32 1, i32 2, i8* [[TMP53]]) 729 // CHECK2-NEXT: store i8* [[TMP54]], i8** [[DOTTASK_RED_]], align 8 730 // CHECK2-NEXT: [[TMP55:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 731 // CHECK2-NEXT: [[TMP56:%.*]] = load i32, i32* [[TMP55]], align 4 732 // CHECK2-NEXT: call void @__kmpc_for_static_init_8(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP56]], i32 34, i32* [[DOTOMP_IS_LAST]], i64* [[DOTOMP_LB]], i64* [[DOTOMP_UB]], i64* [[DOTOMP_STRIDE]], i64 1, i64 1) 733 // CHECK2-NEXT: [[TMP57:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 734 // CHECK2-NEXT: [[CMP:%.*]] = icmp sgt i64 [[TMP57]], 9 735 // CHECK2-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 736 // CHECK2: cond.true: 737 // CHECK2-NEXT: br label [[COND_END:%.*]] 738 // CHECK2: cond.false: 739 // CHECK2-NEXT: [[TMP58:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 740 // CHECK2-NEXT: br label [[COND_END]] 741 // CHECK2: cond.end: 742 // CHECK2-NEXT: [[COND:%.*]] = phi i64 [ 9, [[COND_TRUE]] ], [ [[TMP58]], [[COND_FALSE]] ] 743 // CHECK2-NEXT: store i64 [[COND]], i64* [[DOTOMP_UB]], align 8 744 // CHECK2-NEXT: [[TMP59:%.*]] = load i64, i64* [[DOTOMP_LB]], align 8 745 // CHECK2-NEXT: store i64 [[TMP59]], i64* [[DOTOMP_IV]], align 8 746 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 747 // CHECK2: omp.inner.for.cond: 748 // CHECK2-NEXT: [[TMP60:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 749 // CHECK2-NEXT: [[TMP61:%.*]] = load i64, i64* [[DOTOMP_UB]], align 8 750 // CHECK2-NEXT: [[CMP13:%.*]] = icmp sle i64 [[TMP60]], [[TMP61]] 751 // CHECK2-NEXT: br i1 [[CMP13]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]] 752 // CHECK2: omp.inner.for.cond.cleanup: 753 // CHECK2-NEXT: br label [[OMP_INNER_FOR_END:%.*]] 754 // CHECK2: omp.inner.for.body: 755 // CHECK2-NEXT: [[TMP62:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 756 // CHECK2-NEXT: [[MUL:%.*]] = mul nsw i64 [[TMP62]], 1 757 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i64 0, [[MUL]] 758 // CHECK2-NEXT: store i64 [[ADD]], i64* [[I]], align 8 759 // CHECK2-NEXT: [[TMP63:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 0 760 // CHECK2-NEXT: store i8** [[DOTTASK_RED_]], i8*** [[TMP63]], align 8 761 // CHECK2-NEXT: [[TMP64:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 1 762 // CHECK2-NEXT: store i32* [[ARGC1]], i32** [[TMP64]], align 8 763 // CHECK2-NEXT: [[TMP65:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[AGG_CAPTURED]], i32 0, i32 2 764 // CHECK2-NEXT: [[TMP66:%.*]] = load i8**, i8*** [[_TMP5]], align 8 765 // CHECK2-NEXT: store i8** [[TMP66]], i8*** [[TMP65]], align 8 766 // CHECK2-NEXT: [[TMP67:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 767 // CHECK2-NEXT: [[TMP68:%.*]] = load i32, i32* [[TMP67]], align 4 768 // CHECK2-NEXT: [[TMP69:%.*]] = call i8* @__kmpc_omp_task_alloc(%struct.ident_t* @[[GLOB1]], i32 [[TMP68]], 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*)*)) 769 // CHECK2-NEXT: [[TMP70:%.*]] = bitcast i8* [[TMP69]] to %struct.kmp_task_t_with_privates* 770 // CHECK2-NEXT: [[TMP71:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP70]], i32 0, i32 0 771 // CHECK2-NEXT: [[TMP72:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP71]], i32 0, i32 0 772 // CHECK2-NEXT: [[TMP73:%.*]] = load i8*, i8** [[TMP72]], align 8 773 // CHECK2-NEXT: [[TMP74:%.*]] = bitcast %struct.anon* [[AGG_CAPTURED]] to i8* 774 // CHECK2-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP73]], i8* align 8 [[TMP74]], i64 24, i1 false) 775 // CHECK2-NEXT: [[TMP75:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP70]], i32 0, i32 1 776 // CHECK2-NEXT: [[TMP76:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP75]], i32 0, i32 0 777 // CHECK2-NEXT: [[TMP77:%.*]] = load i8*, i8** [[DOTTASK_RED_]], align 8 778 // CHECK2-NEXT: store i8* [[TMP77]], i8** [[TMP76]], align 8 779 // CHECK2-NEXT: [[TMP78:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 780 // CHECK2-NEXT: [[TMP79:%.*]] = load i32, i32* [[TMP78]], align 4 781 // CHECK2-NEXT: [[TMP80:%.*]] = call i32 @__kmpc_omp_task(%struct.ident_t* @[[GLOB1]], i32 [[TMP79]], i8* [[TMP69]]) 782 // CHECK2-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 783 // CHECK2: omp.body.continue: 784 // CHECK2-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 785 // CHECK2: omp.inner.for.inc: 786 // CHECK2-NEXT: [[TMP81:%.*]] = load i64, i64* [[DOTOMP_IV]], align 8 787 // CHECK2-NEXT: [[ADD14:%.*]] = add nsw i64 [[TMP81]], 1 788 // CHECK2-NEXT: store i64 [[ADD14]], i64* [[DOTOMP_IV]], align 8 789 // CHECK2-NEXT: br label [[OMP_INNER_FOR_COND]] 790 // CHECK2: omp.inner.for.end: 791 // CHECK2-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 792 // CHECK2: omp.loop.exit: 793 // CHECK2-NEXT: [[TMP82:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 794 // CHECK2-NEXT: [[TMP83:%.*]] = load i32, i32* [[TMP82]], align 4 795 // CHECK2-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB2]], i32 [[TMP83]]) 796 // CHECK2-NEXT: [[TMP84:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 797 // CHECK2-NEXT: [[TMP85:%.*]] = load i32, i32* [[TMP84]], align 4 798 // CHECK2-NEXT: call void @__kmpc_task_reduction_modifier_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP85]], i32 1) 799 // CHECK2-NEXT: [[TMP86:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0 800 // CHECK2-NEXT: [[TMP87:%.*]] = bitcast i32* [[ARGC1]] to i8* 801 // CHECK2-NEXT: store i8* [[TMP87]], i8** [[TMP86]], align 8 802 // CHECK2-NEXT: [[TMP88:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1 803 // CHECK2-NEXT: store i8* [[VLA]], i8** [[TMP88]], align 8 804 // CHECK2-NEXT: [[TMP89:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2 805 // CHECK2-NEXT: [[TMP90:%.*]] = inttoptr i64 [[TMP11]] to i8* 806 // CHECK2-NEXT: store i8* [[TMP90]], i8** [[TMP89]], align 8 807 // CHECK2-NEXT: [[TMP91:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 808 // CHECK2-NEXT: [[TMP92:%.*]] = load i32, i32* [[TMP91]], align 4 809 // CHECK2-NEXT: [[TMP93:%.*]] = bitcast [3 x i8*]* [[DOTOMP_REDUCTION_RED_LIST]] to i8* 810 // CHECK2-NEXT: [[TMP94:%.*]] = call i32 @__kmpc_reduce_nowait(%struct.ident_t* @[[GLOB3:[0-9]+]], i32 [[TMP92]], i32 2, i64 24, i8* [[TMP93]], void (i8*, i8*)* @.omp.reduction.reduction_func, [8 x i32]* @.gomp_critical_user_.reduction.var) 811 // CHECK2-NEXT: switch i32 [[TMP94]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [ 812 // CHECK2-NEXT: i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]] 813 // CHECK2-NEXT: i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]] 814 // CHECK2-NEXT: ] 815 // CHECK2: .omp.reduction.case1: 816 // CHECK2-NEXT: [[TMP95:%.*]] = load i32, i32* [[TMP0]], align 4 817 // CHECK2-NEXT: [[TMP96:%.*]] = load i32, i32* [[ARGC1]], align 4 818 // CHECK2-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP95]], [[TMP96]] 819 // CHECK2-NEXT: store i32 [[ADD15]], i32* [[TMP0]], align 4 820 // CHECK2-NEXT: [[TMP97:%.*]] = getelementptr i8, i8* [[ARRAYIDX2]], i64 [[TMP11]] 821 // CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[ARRAYIDX2]], [[TMP97]] 822 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE22:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 823 // CHECK2: omp.arraycpy.body: 824 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 825 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST16:%.*]] = phi i8* [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT20:%.*]], [[OMP_ARRAYCPY_BODY]] ] 826 // CHECK2-NEXT: [[TMP98:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1 827 // CHECK2-NEXT: [[CONV:%.*]] = sext i8 [[TMP98]] to i32 828 // CHECK2-NEXT: [[TMP99:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 829 // CHECK2-NEXT: [[CONV17:%.*]] = sext i8 [[TMP99]] to i32 830 // CHECK2-NEXT: [[ADD18:%.*]] = add nsw i32 [[CONV]], [[CONV17]] 831 // CHECK2-NEXT: [[CONV19:%.*]] = trunc i32 [[ADD18]] to i8 832 // CHECK2-NEXT: store i8 [[CONV19]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST16]], align 1 833 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT20]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST16]], i32 1 834 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 835 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE21:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT20]], [[TMP97]] 836 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_DONE22]], label [[OMP_ARRAYCPY_BODY]] 837 // CHECK2: omp.arraycpy.done22: 838 // CHECK2-NEXT: call void @__kmpc_end_reduce_nowait(%struct.ident_t* @[[GLOB3]], i32 [[TMP92]], [8 x i32]* @.gomp_critical_user_.reduction.var) 839 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 840 // CHECK2: .omp.reduction.case2: 841 // CHECK2-NEXT: [[TMP100:%.*]] = load i32, i32* [[ARGC1]], align 4 842 // CHECK2-NEXT: [[TMP101:%.*]] = atomicrmw add i32* [[TMP0]], i32 [[TMP100]] monotonic, align 4 843 // CHECK2-NEXT: [[TMP102:%.*]] = getelementptr i8, i8* [[ARRAYIDX2]], i64 [[TMP11]] 844 // CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY23:%.*]] = icmp eq i8* [[ARRAYIDX2]], [[TMP102]] 845 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY23]], label [[OMP_ARRAYCPY_DONE36:%.*]], label [[OMP_ARRAYCPY_BODY24:%.*]] 846 // CHECK2: omp.arraycpy.body24: 847 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST25:%.*]] = phi i8* [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT34:%.*]], [[ATOMIC_EXIT:%.*]] ] 848 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST26:%.*]] = phi i8* [ [[ARRAYIDX2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT33:%.*]], [[ATOMIC_EXIT]] ] 849 // CHECK2-NEXT: [[TMP103:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1 850 // CHECK2-NEXT: [[CONV27:%.*]] = sext i8 [[TMP103]] to i32 851 // CHECK2-NEXT: [[ATOMIC_LOAD:%.*]] = load atomic i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST26]] monotonic, align 1 852 // CHECK2-NEXT: br label [[ATOMIC_CONT:%.*]] 853 // CHECK2: atomic_cont: 854 // CHECK2-NEXT: [[TMP104:%.*]] = phi i8 [ [[ATOMIC_LOAD]], [[OMP_ARRAYCPY_BODY24]] ], [ [[TMP109:%.*]], [[ATOMIC_CONT]] ] 855 // CHECK2-NEXT: store i8 [[TMP104]], i8* [[_TMP28]], align 1 856 // CHECK2-NEXT: [[TMP105:%.*]] = load i8, i8* [[_TMP28]], align 1 857 // CHECK2-NEXT: [[CONV29:%.*]] = sext i8 [[TMP105]] to i32 858 // CHECK2-NEXT: [[TMP106:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST25]], align 1 859 // CHECK2-NEXT: [[CONV30:%.*]] = sext i8 [[TMP106]] to i32 860 // CHECK2-NEXT: [[ADD31:%.*]] = add nsw i32 [[CONV29]], [[CONV30]] 861 // CHECK2-NEXT: [[CONV32:%.*]] = trunc i32 [[ADD31]] to i8 862 // CHECK2-NEXT: store i8 [[CONV32]], i8* [[ATOMIC_TEMP]], align 1 863 // CHECK2-NEXT: [[TMP107:%.*]] = load i8, i8* [[ATOMIC_TEMP]], align 1 864 // CHECK2-NEXT: [[TMP108:%.*]] = cmpxchg i8* [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i8 [[TMP104]], i8 [[TMP107]] monotonic monotonic, align 1 865 // CHECK2-NEXT: [[TMP109]] = extractvalue { i8, i1 } [[TMP108]], 0 866 // CHECK2-NEXT: [[TMP110:%.*]] = extractvalue { i8, i1 } [[TMP108]], 1 867 // CHECK2-NEXT: br i1 [[TMP110]], label [[ATOMIC_EXIT]], label [[ATOMIC_CONT]] 868 // CHECK2: atomic_exit: 869 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT33]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i32 1 870 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT34]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST25]], i32 1 871 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE35:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT33]], [[TMP102]] 872 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE35]], label [[OMP_ARRAYCPY_DONE36]], label [[OMP_ARRAYCPY_BODY24]] 873 // CHECK2: omp.arraycpy.done36: 874 // CHECK2-NEXT: br label [[DOTOMP_REDUCTION_DEFAULT]] 875 // CHECK2: .omp.reduction.default: 876 // CHECK2-NEXT: [[TMP111:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 877 // CHECK2-NEXT: call void @llvm.stackrestore(i8* [[TMP111]]) 878 // CHECK2-NEXT: ret void 879 // 880 // 881 // CHECK2-LABEL: define {{[^@]+}}@.red_init. 882 // CHECK2-SAME: (i8* noalias [[TMP0:%.*]], i8* noalias [[TMP1:%.*]]) #[[ATTR3:[0-9]+]] { 883 // CHECK2-NEXT: entry: 884 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 885 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 886 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 887 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 888 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 889 // CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to i32* 890 // CHECK2-NEXT: store i32 0, i32* [[TMP3]], align 8 891 // CHECK2-NEXT: ret void 892 // 893 // 894 // CHECK2-LABEL: define {{[^@]+}}@.red_comb. 895 // CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 896 // CHECK2-NEXT: entry: 897 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 898 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 899 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 900 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 901 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 902 // CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to i32* 903 // CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 904 // CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to i32* 905 // CHECK2-NEXT: [[TMP6:%.*]] = load i32, i32* [[TMP3]], align 8 906 // CHECK2-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP5]], align 8 907 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP6]], [[TMP7]] 908 // CHECK2-NEXT: store i32 [[ADD]], i32* [[TMP3]], align 8 909 // CHECK2-NEXT: ret void 910 // 911 // 912 // CHECK2-LABEL: define {{[^@]+}}@.red_init..1 913 // CHECK2-SAME: (i8* noalias [[TMP0:%.*]], i8* noalias [[TMP1:%.*]]) #[[ATTR3]] { 914 // CHECK2-NEXT: entry: 915 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 916 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 917 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 918 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 919 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 920 // CHECK2-NEXT: [[TMP3:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8 921 // CHECK2-NEXT: [[TMP4:%.*]] = getelementptr i8, i8* [[TMP2]], i64 [[TMP3]] 922 // CHECK2-NEXT: [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq i8* [[TMP2]], [[TMP4]] 923 // CHECK2-NEXT: br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]] 924 // CHECK2: omp.arrayinit.body: 925 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP2]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ] 926 // CHECK2-NEXT: store i8 0, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 927 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 928 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]] 929 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]] 930 // CHECK2: omp.arrayinit.done: 931 // CHECK2-NEXT: ret void 932 // 933 // 934 // CHECK2-LABEL: define {{[^@]+}}@.red_comb..2 935 // CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 936 // CHECK2-NEXT: entry: 937 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 938 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 939 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 940 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 941 // CHECK2-NEXT: [[TMP2:%.*]] = load i64, i64* @{{reduction_size[.].+[.]}}, align 8 942 // CHECK2-NEXT: [[TMP3:%.*]] = load i8*, i8** [[DOTADDR]], align 8 943 // CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 944 // CHECK2-NEXT: [[TMP5:%.*]] = getelementptr i8, i8* [[TMP3]], i64 [[TMP2]] 945 // CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP3]], [[TMP5]] 946 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE4:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 947 // CHECK2: omp.arraycpy.body: 948 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP4]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 949 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP3]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 950 // CHECK2-NEXT: [[TMP6:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 951 // CHECK2-NEXT: [[CONV:%.*]] = sext i8 [[TMP6]] to i32 952 // CHECK2-NEXT: [[TMP7:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 953 // CHECK2-NEXT: [[CONV2:%.*]] = sext i8 [[TMP7]] to i32 954 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 955 // CHECK2-NEXT: [[CONV3:%.*]] = trunc i32 [[ADD]] to i8 956 // CHECK2-NEXT: store i8 [[CONV3]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 957 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 958 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 959 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP5]] 960 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE4]], label [[OMP_ARRAYCPY_BODY]] 961 // CHECK2: omp.arraycpy.done4: 962 // CHECK2-NEXT: ret void 963 // 964 // 965 // CHECK2-LABEL: define {{[^@]+}}@.omp_task_privates_map. 966 // CHECK2-SAME: (%struct..kmp_privates.t* noalias [[TMP0:%.*]], i8*** noalias [[TMP1:%.*]]) #[[ATTR6:[0-9]+]] { 967 // CHECK2-NEXT: entry: 968 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca %struct..kmp_privates.t*, align 8 969 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8***, align 8 970 // CHECK2-NEXT: store %struct..kmp_privates.t* [[TMP0]], %struct..kmp_privates.t** [[DOTADDR]], align 8 971 // CHECK2-NEXT: store i8*** [[TMP1]], i8**** [[DOTADDR1]], align 8 972 // CHECK2-NEXT: [[TMP2:%.*]] = load %struct..kmp_privates.t*, %struct..kmp_privates.t** [[DOTADDR]], align 8 973 // CHECK2-NEXT: [[TMP3:%.*]] = getelementptr inbounds [[STRUCT__KMP_PRIVATES_T:%.*]], %struct..kmp_privates.t* [[TMP2]], i32 0, i32 0 974 // CHECK2-NEXT: [[TMP4:%.*]] = load i8***, i8**** [[DOTADDR1]], align 8 975 // CHECK2-NEXT: store i8** [[TMP3]], i8*** [[TMP4]], align 8 976 // CHECK2-NEXT: ret void 977 // 978 // 979 // CHECK2-LABEL: define {{[^@]+}}@.omp_task_entry. 980 // CHECK2-SAME: (i32 [[TMP0:%.*]], %struct.kmp_task_t_with_privates* noalias [[TMP1:%.*]]) #[[ATTR3]] { 981 // CHECK2-NEXT: entry: 982 // CHECK2-NEXT: [[DOTGLOBAL_TID__ADDR_I:%.*]] = alloca i32, align 4 983 // CHECK2-NEXT: [[DOTPART_ID__ADDR_I:%.*]] = alloca i32*, align 8 984 // CHECK2-NEXT: [[DOTPRIVATES__ADDR_I:%.*]] = alloca i8*, align 8 985 // CHECK2-NEXT: [[DOTCOPY_FN__ADDR_I:%.*]] = alloca void (i8*, ...)*, align 8 986 // CHECK2-NEXT: [[DOTTASK_T__ADDR_I:%.*]] = alloca i8*, align 8 987 // CHECK2-NEXT: [[__CONTEXT_ADDR_I:%.*]] = alloca %struct.anon*, align 8 988 // CHECK2-NEXT: [[DOTFIRSTPRIV_PTR_ADDR_I:%.*]] = alloca i8**, align 8 989 // CHECK2-NEXT: [[TMP_I:%.*]] = alloca i8**, align 8 990 // CHECK2-NEXT: [[TMP4_I:%.*]] = alloca i8*, align 8 991 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i32, align 4 992 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca %struct.kmp_task_t_with_privates*, align 8 993 // CHECK2-NEXT: store i32 [[TMP0]], i32* [[DOTADDR]], align 4 994 // CHECK2-NEXT: store %struct.kmp_task_t_with_privates* [[TMP1]], %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8 995 // CHECK2-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTADDR]], align 4 996 // CHECK2-NEXT: [[TMP3:%.*]] = load %struct.kmp_task_t_with_privates*, %struct.kmp_task_t_with_privates** [[DOTADDR1]], align 8 997 // CHECK2-NEXT: [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES:%.*]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 0 998 // CHECK2-NEXT: [[TMP5:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T:%.*]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 2 999 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T]], %struct.kmp_task_t* [[TMP4]], i32 0, i32 0 1000 // CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 1001 // CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to %struct.anon* 1002 // CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [[STRUCT_KMP_TASK_T_WITH_PRIVATES]], %struct.kmp_task_t_with_privates* [[TMP3]], i32 0, i32 1 1003 // CHECK2-NEXT: [[TMP10:%.*]] = bitcast %struct..kmp_privates.t* [[TMP9]] to i8* 1004 // CHECK2-NEXT: [[TMP11:%.*]] = bitcast %struct.kmp_task_t_with_privates* [[TMP3]] to i8* 1005 // CHECK2-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META3:![0-9]+]]) 1006 // CHECK2-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META6:![0-9]+]]) 1007 // CHECK2-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META8:![0-9]+]]) 1008 // CHECK2-NEXT: call void @llvm.experimental.noalias.scope.decl(metadata [[META10:![0-9]+]]) 1009 // CHECK2-NEXT: store i32 [[TMP2]], i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 1010 // CHECK2-NEXT: store i32* [[TMP5]], i32** [[DOTPART_ID__ADDR_I]], align 8, !noalias !12 1011 // CHECK2-NEXT: store i8* [[TMP10]], i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 1012 // 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 1013 // CHECK2-NEXT: store i8* [[TMP11]], i8** [[DOTTASK_T__ADDR_I]], align 8, !noalias !12 1014 // CHECK2-NEXT: store %struct.anon* [[TMP8]], %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !12 1015 // CHECK2-NEXT: [[TMP12:%.*]] = load %struct.anon*, %struct.anon** [[__CONTEXT_ADDR_I]], align 8, !noalias !12 1016 // CHECK2-NEXT: [[TMP13:%.*]] = load void (i8*, ...)*, void (i8*, ...)** [[DOTCOPY_FN__ADDR_I]], align 8, !noalias !12 1017 // CHECK2-NEXT: [[TMP14:%.*]] = load i8*, i8** [[DOTPRIVATES__ADDR_I]], align 8, !noalias !12 1018 // CHECK2-NEXT: [[TMP15:%.*]] = bitcast void (i8*, ...)* [[TMP13]] to void (i8*, i8***)* 1019 // CHECK2-NEXT: call void [[TMP15]](i8* [[TMP14]], i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]]) #[[ATTR5]] 1020 // CHECK2-NEXT: [[TMP16:%.*]] = load i8**, i8*** [[DOTFIRSTPRIV_PTR_ADDR_I]], align 8, !noalias !12 1021 // CHECK2-NEXT: [[TMP17:%.*]] = getelementptr inbounds [[STRUCT_ANON:%.*]], %struct.anon* [[TMP12]], i32 0, i32 1 1022 // CHECK2-NEXT: [[TMP18:%.*]] = load i32*, i32** [[TMP17]], align 8 1023 // CHECK2-NEXT: [[TMP19:%.*]] = load i8*, i8** [[TMP16]], align 8 1024 // CHECK2-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTGLOBAL_TID__ADDR_I]], align 4, !noalias !12 1025 // CHECK2-NEXT: [[TMP21:%.*]] = bitcast i32* [[TMP18]] to i8* 1026 // CHECK2-NEXT: [[TMP22:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP19]], i8* [[TMP21]]) #[[ATTR5]] 1027 // CHECK2-NEXT: [[CONV_I:%.*]] = bitcast i8* [[TMP22]] to i32* 1028 // CHECK2-NEXT: [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2 1029 // CHECK2-NEXT: [[TMP24:%.*]] = load i8**, i8*** [[TMP23]], align 8 1030 // CHECK2-NEXT: [[TMP25:%.*]] = load i8*, i8** [[TMP24]], align 8 1031 // CHECK2-NEXT: [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 1 1032 // CHECK2-NEXT: [[TMP27:%.*]] = load i32*, i32** [[TMP26]], align 8 1033 // CHECK2-NEXT: [[TMP28:%.*]] = load i32, i32* [[TMP27]], align 4 1034 // CHECK2-NEXT: [[TMP29:%.*]] = sext i32 [[TMP28]] to i64 1035 // CHECK2-NEXT: [[LB_ADD_LEN_I:%.*]] = add nsw i64 -1, [[TMP29]] 1036 // CHECK2-NEXT: [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2 1037 // CHECK2-NEXT: [[TMP31:%.*]] = load i8**, i8*** [[TMP30]], align 8 1038 // CHECK2-NEXT: [[ARRAYIDX2_I:%.*]] = getelementptr inbounds i8*, i8** [[TMP31]], i64 9 1039 // CHECK2-NEXT: [[TMP32:%.*]] = load i8*, i8** [[ARRAYIDX2_I]], align 8 1040 // CHECK2-NEXT: [[ARRAYIDX3_I:%.*]] = getelementptr inbounds i8, i8* [[TMP32]], i64 [[LB_ADD_LEN_I]] 1041 // CHECK2-NEXT: [[TMP33:%.*]] = ptrtoint i8* [[ARRAYIDX3_I]] to i64 1042 // CHECK2-NEXT: [[TMP34:%.*]] = ptrtoint i8* [[TMP25]] to i64 1043 // CHECK2-NEXT: [[TMP35:%.*]] = sub i64 [[TMP33]], [[TMP34]] 1044 // CHECK2-NEXT: [[TMP36:%.*]] = sdiv exact i64 [[TMP35]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 1045 // CHECK2-NEXT: [[TMP37:%.*]] = add nuw i64 [[TMP36]], 1 1046 // CHECK2-NEXT: [[TMP38:%.*]] = mul nuw i64 [[TMP37]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 1047 // CHECK2-NEXT: store i64 [[TMP37]], i64* @{{reduction_size[.].+[.]}}, align 8 1048 // CHECK2-NEXT: [[TMP39:%.*]] = load i8*, i8** [[TMP16]], align 8 1049 // CHECK2-NEXT: [[TMP40:%.*]] = call i8* @__kmpc_task_reduction_get_th_data(i32 [[TMP20]], i8* [[TMP39]], i8* [[TMP25]]) #[[ATTR5]] 1050 // CHECK2-NEXT: [[TMP41:%.*]] = getelementptr inbounds [[STRUCT_ANON]], %struct.anon* [[TMP12]], i32 0, i32 2 1051 // CHECK2-NEXT: [[TMP42:%.*]] = load i8**, i8*** [[TMP41]], align 8 1052 // CHECK2-NEXT: [[TMP43:%.*]] = load i8*, i8** [[TMP42]], align 8 1053 // CHECK2-NEXT: [[TMP44:%.*]] = ptrtoint i8* [[TMP43]] to i64 1054 // CHECK2-NEXT: [[TMP45:%.*]] = ptrtoint i8* [[TMP25]] to i64 1055 // CHECK2-NEXT: [[TMP46:%.*]] = sub i64 [[TMP44]], [[TMP45]] 1056 // CHECK2-NEXT: [[TMP47:%.*]] = sdiv exact i64 [[TMP46]], ptrtoint (i8* getelementptr (i8, i8* null, i32 1) to i64) 1057 // CHECK2-NEXT: [[TMP48:%.*]] = getelementptr i8, i8* [[TMP40]], i64 [[TMP47]] 1058 // CHECK2-NEXT: store i8** [[TMP4_I]], i8*** [[TMP_I]], align 8, !noalias !12 1059 // CHECK2-NEXT: store i8* [[TMP48]], i8** [[TMP4_I]], align 8, !noalias !12 1060 // CHECK2-NEXT: ret i32 0 1061 // 1062 // 1063 // CHECK2-LABEL: define {{[^@]+}}@.omp.reduction.reduction_func 1064 // CHECK2-SAME: (i8* [[TMP0:%.*]], i8* [[TMP1:%.*]]) #[[ATTR3]] { 1065 // CHECK2-NEXT: entry: 1066 // CHECK2-NEXT: [[DOTADDR:%.*]] = alloca i8*, align 8 1067 // CHECK2-NEXT: [[DOTADDR1:%.*]] = alloca i8*, align 8 1068 // CHECK2-NEXT: store i8* [[TMP0]], i8** [[DOTADDR]], align 8 1069 // CHECK2-NEXT: store i8* [[TMP1]], i8** [[DOTADDR1]], align 8 1070 // CHECK2-NEXT: [[TMP2:%.*]] = load i8*, i8** [[DOTADDR]], align 8 1071 // CHECK2-NEXT: [[TMP3:%.*]] = bitcast i8* [[TMP2]] to [3 x i8*]* 1072 // CHECK2-NEXT: [[TMP4:%.*]] = load i8*, i8** [[DOTADDR1]], align 8 1073 // CHECK2-NEXT: [[TMP5:%.*]] = bitcast i8* [[TMP4]] to [3 x i8*]* 1074 // CHECK2-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 0 1075 // CHECK2-NEXT: [[TMP7:%.*]] = load i8*, i8** [[TMP6]], align 8 1076 // CHECK2-NEXT: [[TMP8:%.*]] = bitcast i8* [[TMP7]] to i32* 1077 // CHECK2-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 0 1078 // CHECK2-NEXT: [[TMP10:%.*]] = load i8*, i8** [[TMP9]], align 8 1079 // CHECK2-NEXT: [[TMP11:%.*]] = bitcast i8* [[TMP10]] to i32* 1080 // CHECK2-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP5]], i64 0, i64 1 1081 // CHECK2-NEXT: [[TMP13:%.*]] = load i8*, i8** [[TMP12]], align 8 1082 // CHECK2-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 1 1083 // CHECK2-NEXT: [[TMP15:%.*]] = load i8*, i8** [[TMP14]], align 8 1084 // CHECK2-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[TMP3]], i64 0, i64 2 1085 // CHECK2-NEXT: [[TMP17:%.*]] = load i8*, i8** [[TMP16]], align 8 1086 // CHECK2-NEXT: [[TMP18:%.*]] = ptrtoint i8* [[TMP17]] to i64 1087 // CHECK2-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP11]], align 4 1088 // CHECK2-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP8]], align 4 1089 // CHECK2-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 1090 // CHECK2-NEXT: store i32 [[ADD]], i32* [[TMP11]], align 4 1091 // CHECK2-NEXT: [[TMP21:%.*]] = getelementptr i8, i8* [[TMP15]], i64 [[TMP18]] 1092 // CHECK2-NEXT: [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq i8* [[TMP15]], [[TMP21]] 1093 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE5:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]] 1094 // CHECK2: omp.arraycpy.body: 1095 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi i8* [ [[TMP13]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 1096 // CHECK2-NEXT: [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi i8* [ [[TMP15]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ] 1097 // CHECK2-NEXT: [[TMP22:%.*]] = load i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 1098 // CHECK2-NEXT: [[CONV:%.*]] = sext i8 [[TMP22]] to i32 1099 // CHECK2-NEXT: [[TMP23:%.*]] = load i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 1 1100 // CHECK2-NEXT: [[CONV2:%.*]] = sext i8 [[TMP23]] to i32 1101 // CHECK2-NEXT: [[ADD3:%.*]] = add nsw i32 [[CONV]], [[CONV2]] 1102 // CHECK2-NEXT: [[CONV4:%.*]] = trunc i32 [[ADD3]] to i8 1103 // CHECK2-NEXT: store i8 [[CONV4]], i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 1 1104 // CHECK2-NEXT: [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1 1105 // CHECK2-NEXT: [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i8, i8* [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1 1106 // CHECK2-NEXT: [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq i8* [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP21]] 1107 // CHECK2-NEXT: br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE5]], label [[OMP_ARRAYCPY_BODY]] 1108 // CHECK2: omp.arraycpy.done5: 1109 // CHECK2-NEXT: ret void 1110 // 1111