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 // expected-no-diagnostics 3 #ifndef HEADER 4 #define HEADER 5 // Test host codegen. 6 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1 7 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 8 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1 9 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3 10 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 11 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK3 12 13 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 14 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 15 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 16 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 17 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 18 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 19 #ifdef CK1 20 21 int a[100]; 22 23 int teams_argument_global(int n) { 24 int te = n / 128; 25 int th = 128; 26 // discard n_addr 27 28 29 #pragma omp target 30 #pragma omp teams distribute num_teams(te), thread_limit(th) 31 for(int i = 0; i < n; i++) { 32 a[i] = 0; 33 } 34 35 #pragma omp target 36 {{{ 37 #pragma omp teams distribute 38 for(int i = 0; i < n; i++) { 39 a[i] = 0; 40 } 41 }}} 42 43 // outlined target regions 44 45 46 47 48 return a[0]; 49 } 50 51 #endif // CK1 52 53 // Test host codegen. 54 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK9 55 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 56 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK9 57 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK11 58 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 59 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK11 60 61 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 62 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 63 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 64 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 65 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 66 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 67 #ifdef CK2 68 69 int teams_local_arg(void) { 70 int n = 100; 71 int a[n]; 72 73 #pragma omp target 74 #pragma omp teams distribute 75 for(int i = 0; i < n; i++) { 76 a[i] = 0; 77 } 78 79 // outlined target region 80 81 82 return a[0]; 83 } 84 #endif // CK2 85 86 // Test host codegen. 87 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK17 88 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 89 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK17 90 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK19 91 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 92 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK19 93 94 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 95 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 96 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 97 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 98 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 99 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 100 #ifdef CK3 101 102 103 template <typename T, int X, long long Y> 104 struct SS{ 105 T a[X]; 106 float b; 107 int foo(void) { 108 109 #pragma omp target 110 #pragma omp teams distribute 111 for(int i = 0; i < X; i++) { 112 a[i] = (T)0; 113 } 114 115 // outlined target region 116 117 118 return a[0]; 119 } 120 }; 121 122 int teams_template_struct(void) { 123 SS<int, 123, 456> V; 124 return V.foo(); 125 126 } 127 #endif // CK3 128 129 // Test host codegen. 130 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK25 131 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 132 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK25 133 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK27 134 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 135 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK27 136 137 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 138 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 139 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 140 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 141 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 142 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 143 144 #ifdef CK4 145 146 template <typename T, int n> 147 int tmain(T argc) { 148 T a[n]; 149 int te = n/128; 150 int th = 128; 151 #pragma omp target 152 #pragma omp teams distribute num_teams(te) thread_limit(th) 153 for(int i = 0; i < n; i++) { 154 a[i] = (T)0; 155 } 156 return 0; 157 } 158 159 int main (int argc, char **argv) { 160 int n = 100; 161 int a[n]; 162 #pragma omp target 163 #pragma omp teams distribute 164 for(int i = 0; i < n; i++) { 165 a[i] = 0; 166 } 167 return tmain<int, 10>(argc); 168 } 169 170 171 172 173 174 175 176 #endif // CK4 177 #endif 178 // CHECK1-LABEL: define {{[^@]+}}@_Z21teams_argument_globali 179 // CHECK1-SAME: (i32 noundef signext [[N:%.*]]) #[[ATTR0:[0-9]+]] { 180 // CHECK1-NEXT: entry: 181 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 182 // CHECK1-NEXT: [[TE:%.*]] = alloca i32, align 4 183 // CHECK1-NEXT: [[TH:%.*]] = alloca i32, align 4 184 // CHECK1-NEXT: [[TE_CASTED:%.*]] = alloca i64, align 8 185 // CHECK1-NEXT: [[TH_CASTED:%.*]] = alloca i64, align 8 186 // CHECK1-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 187 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 8 188 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8 189 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8 190 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 191 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 192 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 193 // CHECK1-NEXT: [[N_CASTED6:%.*]] = alloca i64, align 8 194 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS8:%.*]] = alloca [2 x i8*], align 8 195 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS9:%.*]] = alloca [2 x i8*], align 8 196 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS10:%.*]] = alloca [2 x i8*], align 8 197 // CHECK1-NEXT: [[_TMP11:%.*]] = alloca i32, align 4 198 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_12:%.*]] = alloca i32, align 4 199 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_13:%.*]] = alloca i32, align 4 200 // CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 201 // CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 202 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP0]], 128 203 // CHECK1-NEXT: store i32 [[DIV]], i32* [[TE]], align 4 204 // CHECK1-NEXT: store i32 128, i32* [[TH]], align 4 205 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TE]], align 4 206 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[TE_CASTED]] to i32* 207 // CHECK1-NEXT: store i32 [[TMP1]], i32* [[CONV]], align 4 208 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[TE_CASTED]], align 8 209 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TH]], align 4 210 // CHECK1-NEXT: [[CONV1:%.*]] = bitcast i64* [[TH_CASTED]] to i32* 211 // CHECK1-NEXT: store i32 [[TMP3]], i32* [[CONV1]], align 4 212 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[TH_CASTED]], align 8 213 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[N_ADDR]], align 4 214 // CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32* 215 // CHECK1-NEXT: store i32 [[TMP5]], i32* [[CONV2]], align 4 216 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, i64* [[N_CASTED]], align 8 217 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 218 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64* 219 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP8]], align 8 220 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 221 // CHECK1-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64* 222 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP10]], align 8 223 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 224 // CHECK1-NEXT: store i8* null, i8** [[TMP11]], align 8 225 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 226 // CHECK1-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* 227 // CHECK1-NEXT: store i64 [[TMP4]], i64* [[TMP13]], align 8 228 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 229 // CHECK1-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* 230 // CHECK1-NEXT: store i64 [[TMP4]], i64* [[TMP15]], align 8 231 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 232 // CHECK1-NEXT: store i8* null, i8** [[TMP16]], align 8 233 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 234 // CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64* 235 // CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8 236 // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 237 // CHECK1-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64* 238 // CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP20]], align 8 239 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 240 // CHECK1-NEXT: store i8* null, i8** [[TMP21]], align 8 241 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 242 // CHECK1-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to [100 x i32]** 243 // CHECK1-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP23]], align 8 244 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 245 // CHECK1-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [100 x i32]** 246 // CHECK1-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP25]], align 8 247 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 248 // CHECK1-NEXT: store i8* null, i8** [[TMP26]], align 8 249 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 250 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 251 // CHECK1-NEXT: [[TMP29:%.*]] = load i32, i32* [[TE]], align 4 252 // CHECK1-NEXT: [[TMP30:%.*]] = load i32, i32* [[TH]], align 4 253 // CHECK1-NEXT: [[TMP31:%.*]] = load i32, i32* [[N_ADDR]], align 4 254 // CHECK1-NEXT: store i32 [[TMP31]], i32* [[DOTCAPTURE_EXPR_]], align 4 255 // CHECK1-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 256 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP32]], 0 257 // CHECK1-NEXT: [[DIV4:%.*]] = sdiv i32 [[SUB]], 1 258 // CHECK1-NEXT: [[SUB5:%.*]] = sub nsw i32 [[DIV4]], 1 259 // CHECK1-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4 260 // CHECK1-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 261 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP33]], 1 262 // CHECK1-NEXT: [[TMP34:%.*]] = zext i32 [[ADD]] to i64 263 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP34]]) 264 // CHECK1-NEXT: [[TMP35:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l29.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 [[TMP29]], i32 [[TMP30]]) 265 // CHECK1-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0 266 // CHECK1-NEXT: br i1 [[TMP36]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 267 // CHECK1: omp_offload.failed: 268 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l29(i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [100 x i32]* @a) #[[ATTR2:[0-9]+]] 269 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] 270 // CHECK1: omp_offload.cont: 271 // CHECK1-NEXT: [[TMP37:%.*]] = load i32, i32* [[N_ADDR]], align 4 272 // CHECK1-NEXT: [[CONV7:%.*]] = bitcast i64* [[N_CASTED6]] to i32* 273 // CHECK1-NEXT: store i32 [[TMP37]], i32* [[CONV7]], align 4 274 // CHECK1-NEXT: [[TMP38:%.*]] = load i64, i64* [[N_CASTED6]], align 8 275 // CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0 276 // CHECK1-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i64* 277 // CHECK1-NEXT: store i64 [[TMP38]], i64* [[TMP40]], align 8 278 // CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS9]], i32 0, i32 0 279 // CHECK1-NEXT: [[TMP42:%.*]] = bitcast i8** [[TMP41]] to i64* 280 // CHECK1-NEXT: store i64 [[TMP38]], i64* [[TMP42]], align 8 281 // CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS10]], i64 0, i64 0 282 // CHECK1-NEXT: store i8* null, i8** [[TMP43]], align 8 283 // CHECK1-NEXT: [[TMP44:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 1 284 // CHECK1-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to [100 x i32]** 285 // CHECK1-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP45]], align 8 286 // CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS9]], i32 0, i32 1 287 // CHECK1-NEXT: [[TMP47:%.*]] = bitcast i8** [[TMP46]] to [100 x i32]** 288 // CHECK1-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP47]], align 8 289 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS10]], i64 0, i64 1 290 // CHECK1-NEXT: store i8* null, i8** [[TMP48]], align 8 291 // CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0 292 // CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS9]], i32 0, i32 0 293 // CHECK1-NEXT: [[TMP51:%.*]] = load i32, i32* [[N_ADDR]], align 4 294 // CHECK1-NEXT: store i32 [[TMP51]], i32* [[DOTCAPTURE_EXPR_12]], align 4 295 // CHECK1-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_12]], align 4 296 // CHECK1-NEXT: [[SUB14:%.*]] = sub nsw i32 [[TMP52]], 0 297 // CHECK1-NEXT: [[DIV15:%.*]] = sdiv i32 [[SUB14]], 1 298 // CHECK1-NEXT: [[SUB16:%.*]] = sub nsw i32 [[DIV15]], 1 299 // CHECK1-NEXT: store i32 [[SUB16]], i32* [[DOTCAPTURE_EXPR_13]], align 4 300 // CHECK1-NEXT: [[TMP53:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_13]], align 4 301 // CHECK1-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP53]], 1 302 // CHECK1-NEXT: [[TMP54:%.*]] = zext i32 [[ADD17]] to i64 303 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP54]]) 304 // CHECK1-NEXT: [[TMP55:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35.region_id, i32 2, i8** [[TMP49]], i8** [[TMP50]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) 305 // CHECK1-NEXT: [[TMP56:%.*]] = icmp ne i32 [[TMP55]], 0 306 // CHECK1-NEXT: br i1 [[TMP56]], label [[OMP_OFFLOAD_FAILED18:%.*]], label [[OMP_OFFLOAD_CONT19:%.*]] 307 // CHECK1: omp_offload.failed18: 308 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35(i64 [[TMP38]], [100 x i32]* @a) #[[ATTR2]] 309 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT19]] 310 // CHECK1: omp_offload.cont19: 311 // CHECK1-NEXT: [[TMP57:%.*]] = load i32, i32* getelementptr inbounds ([100 x i32], [100 x i32]* @a, i64 0, i64 0), align 4 312 // CHECK1-NEXT: ret i32 [[TMP57]] 313 // 314 // 315 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l29 316 // CHECK1-SAME: (i64 noundef [[TE:%.*]], i64 noundef [[TH:%.*]], i64 noundef [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1:[0-9]+]] { 317 // CHECK1-NEXT: entry: 318 // CHECK1-NEXT: [[TE_ADDR:%.*]] = alloca i64, align 8 319 // CHECK1-NEXT: [[TH_ADDR:%.*]] = alloca i64, align 8 320 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 321 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 8 322 // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) 323 // CHECK1-NEXT: store i64 [[TE]], i64* [[TE_ADDR]], align 8 324 // CHECK1-NEXT: store i64 [[TH]], i64* [[TH_ADDR]], align 8 325 // CHECK1-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 326 // CHECK1-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 8 327 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[TE_ADDR]] to i32* 328 // CHECK1-NEXT: [[CONV1:%.*]] = bitcast i64* [[TH_ADDR]] to i32* 329 // CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_ADDR]] to i32* 330 // CHECK1-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 8 331 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 332 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[CONV1]], align 4 333 // CHECK1-NEXT: call void @__kmpc_push_num_teams(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 334 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, [100 x i32]*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[CONV2]], [100 x i32]* [[TMP1]]) 335 // CHECK1-NEXT: ret void 336 // 337 // 338 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined. 339 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 340 // CHECK1-NEXT: entry: 341 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 342 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 343 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 344 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 8 345 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 346 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 347 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 348 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 349 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 350 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 351 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 352 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 353 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 354 // CHECK1-NEXT: [[I3:%.*]] = alloca i32, align 4 355 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 356 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 357 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 358 // CHECK1-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 8 359 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 360 // CHECK1-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 8 361 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP0]], align 4 362 // CHECK1-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 363 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 364 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 365 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 366 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 367 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 368 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 369 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 370 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 371 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 372 // CHECK1: omp.precond.then: 373 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 374 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 375 // CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4 376 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 377 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 378 // CHECK1-NEXT: [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 379 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4 380 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP7]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 381 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 382 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 383 // CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 384 // CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 385 // CHECK1: cond.true: 386 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 387 // CHECK1-NEXT: br label [[COND_END:%.*]] 388 // CHECK1: cond.false: 389 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 390 // CHECK1-NEXT: br label [[COND_END]] 391 // CHECK1: cond.end: 392 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 393 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 394 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 395 // CHECK1-NEXT: store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4 396 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 397 // CHECK1: omp.inner.for.cond: 398 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 399 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 400 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 401 // CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 402 // CHECK1: omp.inner.for.body: 403 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 404 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP15]], 1 405 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 406 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I3]], align 4 407 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[I3]], align 4 408 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP16]] to i64 409 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* [[TMP1]], i64 0, i64 [[IDXPROM]] 410 // CHECK1-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 411 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 412 // CHECK1: omp.body.continue: 413 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 414 // CHECK1: omp.inner.for.inc: 415 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 416 // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP17]], 1 417 // CHECK1-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4 418 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 419 // CHECK1: omp.inner.for.end: 420 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 421 // CHECK1: omp.loop.exit: 422 // CHECK1-NEXT: [[TMP18:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 423 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 4 424 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP19]]) 425 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 426 // CHECK1: omp.precond.end: 427 // CHECK1-NEXT: ret void 428 // 429 // 430 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35 431 // CHECK1-SAME: (i64 noundef [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 432 // CHECK1-NEXT: entry: 433 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 434 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 8 435 // CHECK1-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 436 // CHECK1-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 8 437 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 438 // CHECK1-NEXT: [[TMP0:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 8 439 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, [100 x i32]*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32* [[CONV]], [100 x i32]* [[TMP0]]) 440 // CHECK1-NEXT: ret void 441 // 442 // 443 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1 444 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 445 // CHECK1-NEXT: entry: 446 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 447 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 448 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 449 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 8 450 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 451 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 452 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 453 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 454 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 455 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 456 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 457 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 458 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 459 // CHECK1-NEXT: [[I3:%.*]] = alloca i32, align 4 460 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 461 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 462 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 463 // CHECK1-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 8 464 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 465 // CHECK1-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 8 466 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP0]], align 4 467 // CHECK1-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 468 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 469 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 470 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 471 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 472 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 473 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 474 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 475 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 476 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 477 // CHECK1: omp.precond.then: 478 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 479 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 480 // CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4 481 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 482 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 483 // CHECK1-NEXT: [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 484 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4 485 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 486 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 487 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 488 // CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 489 // CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 490 // CHECK1: cond.true: 491 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 492 // CHECK1-NEXT: br label [[COND_END:%.*]] 493 // CHECK1: cond.false: 494 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 495 // CHECK1-NEXT: br label [[COND_END]] 496 // CHECK1: cond.end: 497 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 498 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 499 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 500 // CHECK1-NEXT: store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4 501 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 502 // CHECK1: omp.inner.for.cond: 503 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 504 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 505 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 506 // CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 507 // CHECK1: omp.inner.for.body: 508 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 509 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP15]], 1 510 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 511 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I3]], align 4 512 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[I3]], align 4 513 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP16]] to i64 514 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* [[TMP1]], i64 0, i64 [[IDXPROM]] 515 // CHECK1-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 516 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 517 // CHECK1: omp.body.continue: 518 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 519 // CHECK1: omp.inner.for.inc: 520 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 521 // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP17]], 1 522 // CHECK1-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4 523 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 524 // CHECK1: omp.inner.for.end: 525 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 526 // CHECK1: omp.loop.exit: 527 // CHECK1-NEXT: [[TMP18:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 528 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 4 529 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP19]]) 530 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 531 // CHECK1: omp.precond.end: 532 // CHECK1-NEXT: ret void 533 // 534 // 535 // CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 536 // CHECK1-SAME: () #[[ATTR3:[0-9]+]] { 537 // CHECK1-NEXT: entry: 538 // CHECK1-NEXT: call void @__tgt_register_requires(i64 1) 539 // CHECK1-NEXT: ret void 540 // 541 // 542 // CHECK3-LABEL: define {{[^@]+}}@_Z21teams_argument_globali 543 // CHECK3-SAME: (i32 noundef [[N:%.*]]) #[[ATTR0:[0-9]+]] { 544 // CHECK3-NEXT: entry: 545 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 546 // CHECK3-NEXT: [[TE:%.*]] = alloca i32, align 4 547 // CHECK3-NEXT: [[TH:%.*]] = alloca i32, align 4 548 // CHECK3-NEXT: [[TE_CASTED:%.*]] = alloca i32, align 4 549 // CHECK3-NEXT: [[TH_CASTED:%.*]] = alloca i32, align 4 550 // CHECK3-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 551 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4 552 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4 553 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4 554 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 555 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 556 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 557 // CHECK3-NEXT: [[N_CASTED4:%.*]] = alloca i32, align 4 558 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [2 x i8*], align 4 559 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [2 x i8*], align 4 560 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [2 x i8*], align 4 561 // CHECK3-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 562 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 563 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 564 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 565 // CHECK3-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 566 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP0]], 128 567 // CHECK3-NEXT: store i32 [[DIV]], i32* [[TE]], align 4 568 // CHECK3-NEXT: store i32 128, i32* [[TH]], align 4 569 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[TE]], align 4 570 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TE_CASTED]], align 4 571 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[TE_CASTED]], align 4 572 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[TH]], align 4 573 // CHECK3-NEXT: store i32 [[TMP3]], i32* [[TH_CASTED]], align 4 574 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[TH_CASTED]], align 4 575 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[N_ADDR]], align 4 576 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[N_CASTED]], align 4 577 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[N_CASTED]], align 4 578 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 579 // CHECK3-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i32* 580 // CHECK3-NEXT: store i32 [[TMP2]], i32* [[TMP8]], align 4 581 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 582 // CHECK3-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32* 583 // CHECK3-NEXT: store i32 [[TMP2]], i32* [[TMP10]], align 4 584 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 585 // CHECK3-NEXT: store i8* null, i8** [[TMP11]], align 4 586 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 587 // CHECK3-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* 588 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[TMP13]], align 4 589 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 590 // CHECK3-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* 591 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[TMP15]], align 4 592 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 593 // CHECK3-NEXT: store i8* null, i8** [[TMP16]], align 4 594 // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 595 // CHECK3-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* 596 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[TMP18]], align 4 597 // CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 598 // CHECK3-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* 599 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[TMP20]], align 4 600 // CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 601 // CHECK3-NEXT: store i8* null, i8** [[TMP21]], align 4 602 // CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 603 // CHECK3-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to [100 x i32]** 604 // CHECK3-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP23]], align 4 605 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 606 // CHECK3-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [100 x i32]** 607 // CHECK3-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP25]], align 4 608 // CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 609 // CHECK3-NEXT: store i8* null, i8** [[TMP26]], align 4 610 // CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 611 // CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 612 // CHECK3-NEXT: [[TMP29:%.*]] = load i32, i32* [[TE]], align 4 613 // CHECK3-NEXT: [[TMP30:%.*]] = load i32, i32* [[TH]], align 4 614 // CHECK3-NEXT: [[TMP31:%.*]] = load i32, i32* [[N_ADDR]], align 4 615 // CHECK3-NEXT: store i32 [[TMP31]], i32* [[DOTCAPTURE_EXPR_]], align 4 616 // CHECK3-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 617 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP32]], 0 618 // CHECK3-NEXT: [[DIV2:%.*]] = sdiv i32 [[SUB]], 1 619 // CHECK3-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV2]], 1 620 // CHECK3-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_1]], align 4 621 // CHECK3-NEXT: [[TMP33:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 622 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP33]], 1 623 // CHECK3-NEXT: [[TMP34:%.*]] = zext i32 [[ADD]] to i64 624 // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP34]]) 625 // CHECK3-NEXT: [[TMP35:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l29.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 [[TMP29]], i32 [[TMP30]]) 626 // CHECK3-NEXT: [[TMP36:%.*]] = icmp ne i32 [[TMP35]], 0 627 // CHECK3-NEXT: br i1 [[TMP36]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 628 // CHECK3: omp_offload.failed: 629 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l29(i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [100 x i32]* @a) #[[ATTR2:[0-9]+]] 630 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] 631 // CHECK3: omp_offload.cont: 632 // CHECK3-NEXT: [[TMP37:%.*]] = load i32, i32* [[N_ADDR]], align 4 633 // CHECK3-NEXT: store i32 [[TMP37]], i32* [[N_CASTED4]], align 4 634 // CHECK3-NEXT: [[TMP38:%.*]] = load i32, i32* [[N_CASTED4]], align 4 635 // CHECK3-NEXT: [[TMP39:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 636 // CHECK3-NEXT: [[TMP40:%.*]] = bitcast i8** [[TMP39]] to i32* 637 // CHECK3-NEXT: store i32 [[TMP38]], i32* [[TMP40]], align 4 638 // CHECK3-NEXT: [[TMP41:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 639 // CHECK3-NEXT: [[TMP42:%.*]] = bitcast i8** [[TMP41]] to i32* 640 // CHECK3-NEXT: store i32 [[TMP38]], i32* [[TMP42]], align 4 641 // CHECK3-NEXT: [[TMP43:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i32 0, i32 0 642 // CHECK3-NEXT: store i8* null, i8** [[TMP43]], align 4 643 // CHECK3-NEXT: [[TMP44:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 644 // CHECK3-NEXT: [[TMP45:%.*]] = bitcast i8** [[TMP44]] to [100 x i32]** 645 // CHECK3-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP45]], align 4 646 // CHECK3-NEXT: [[TMP46:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 647 // CHECK3-NEXT: [[TMP47:%.*]] = bitcast i8** [[TMP46]] to [100 x i32]** 648 // CHECK3-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP47]], align 4 649 // CHECK3-NEXT: [[TMP48:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i32 0, i32 1 650 // CHECK3-NEXT: store i8* null, i8** [[TMP48]], align 4 651 // CHECK3-NEXT: [[TMP49:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 652 // CHECK3-NEXT: [[TMP50:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 653 // CHECK3-NEXT: [[TMP51:%.*]] = load i32, i32* [[N_ADDR]], align 4 654 // CHECK3-NEXT: store i32 [[TMP51]], i32* [[DOTCAPTURE_EXPR_9]], align 4 655 // CHECK3-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 656 // CHECK3-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP52]], 0 657 // CHECK3-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 658 // CHECK3-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 659 // CHECK3-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 660 // CHECK3-NEXT: [[TMP53:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 661 // CHECK3-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP53]], 1 662 // CHECK3-NEXT: [[TMP54:%.*]] = zext i32 [[ADD14]] to i64 663 // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 [[TMP54]]) 664 // CHECK3-NEXT: [[TMP55:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35.region_id, i32 2, i8** [[TMP49]], i8** [[TMP50]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) 665 // CHECK3-NEXT: [[TMP56:%.*]] = icmp ne i32 [[TMP55]], 0 666 // CHECK3-NEXT: br i1 [[TMP56]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] 667 // CHECK3: omp_offload.failed15: 668 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35(i32 [[TMP38]], [100 x i32]* @a) #[[ATTR2]] 669 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT16]] 670 // CHECK3: omp_offload.cont16: 671 // CHECK3-NEXT: [[TMP57:%.*]] = load i32, i32* getelementptr inbounds ([100 x i32], [100 x i32]* @a, i32 0, i32 0), align 4 672 // CHECK3-NEXT: ret i32 [[TMP57]] 673 // 674 // 675 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l29 676 // CHECK3-SAME: (i32 noundef [[TE:%.*]], i32 noundef [[TH:%.*]], i32 noundef [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1:[0-9]+]] { 677 // CHECK3-NEXT: entry: 678 // CHECK3-NEXT: [[TE_ADDR:%.*]] = alloca i32, align 4 679 // CHECK3-NEXT: [[TH_ADDR:%.*]] = alloca i32, align 4 680 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 681 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 4 682 // CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) 683 // CHECK3-NEXT: store i32 [[TE]], i32* [[TE_ADDR]], align 4 684 // CHECK3-NEXT: store i32 [[TH]], i32* [[TH_ADDR]], align 4 685 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 686 // CHECK3-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 4 687 // CHECK3-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 4 688 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[TE_ADDR]], align 4 689 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[TH_ADDR]], align 4 690 // CHECK3-NEXT: call void @__kmpc_push_num_teams(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 691 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, [100 x i32]*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[N_ADDR]], [100 x i32]* [[TMP1]]) 692 // CHECK3-NEXT: ret void 693 // 694 // 695 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined. 696 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 697 // CHECK3-NEXT: entry: 698 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 699 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 700 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 701 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 4 702 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 703 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 704 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 705 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 706 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 707 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 708 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 709 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 710 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 711 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 712 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 713 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 714 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 715 // CHECK3-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 4 716 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 717 // CHECK3-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 4 718 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP0]], align 4 719 // CHECK3-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 720 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 721 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 722 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 723 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 724 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 725 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 726 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 727 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 728 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 729 // CHECK3: omp.precond.then: 730 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 731 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 732 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4 733 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 734 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 735 // CHECK3-NEXT: [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 736 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4 737 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP7]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 738 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 739 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 740 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 741 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 742 // CHECK3: cond.true: 743 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 744 // CHECK3-NEXT: br label [[COND_END:%.*]] 745 // CHECK3: cond.false: 746 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 747 // CHECK3-NEXT: br label [[COND_END]] 748 // CHECK3: cond.end: 749 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 750 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 751 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 752 // CHECK3-NEXT: store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4 753 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 754 // CHECK3: omp.inner.for.cond: 755 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 756 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 757 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 758 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 759 // CHECK3: omp.inner.for.body: 760 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 761 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP15]], 1 762 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 763 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I3]], align 4 764 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[I3]], align 4 765 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* [[TMP1]], i32 0, i32 [[TMP16]] 766 // CHECK3-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 767 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 768 // CHECK3: omp.body.continue: 769 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 770 // CHECK3: omp.inner.for.inc: 771 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 772 // CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP17]], 1 773 // CHECK3-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4 774 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 775 // CHECK3: omp.inner.for.end: 776 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 777 // CHECK3: omp.loop.exit: 778 // CHECK3-NEXT: [[TMP18:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 779 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 4 780 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP19]]) 781 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 782 // CHECK3: omp.precond.end: 783 // CHECK3-NEXT: ret void 784 // 785 // 786 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35 787 // CHECK3-SAME: (i32 noundef [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 788 // CHECK3-NEXT: entry: 789 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 790 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 4 791 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 792 // CHECK3-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 4 793 // CHECK3-NEXT: [[TMP0:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 4 794 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, [100 x i32]*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], [100 x i32]* [[TMP0]]) 795 // CHECK3-NEXT: ret void 796 // 797 // 798 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..1 799 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 800 // CHECK3-NEXT: entry: 801 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 802 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 803 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 804 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 4 805 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 806 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 807 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 808 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 809 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 810 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 811 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 812 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 813 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 814 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 815 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 816 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 817 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 818 // CHECK3-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 4 819 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 820 // CHECK3-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 4 821 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP0]], align 4 822 // CHECK3-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 823 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 824 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 825 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 826 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 827 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 828 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 829 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 830 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 831 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 832 // CHECK3: omp.precond.then: 833 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 834 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 835 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4 836 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 837 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 838 // CHECK3-NEXT: [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 839 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4 840 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 841 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 842 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 843 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 844 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 845 // CHECK3: cond.true: 846 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 847 // CHECK3-NEXT: br label [[COND_END:%.*]] 848 // CHECK3: cond.false: 849 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 850 // CHECK3-NEXT: br label [[COND_END]] 851 // CHECK3: cond.end: 852 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 853 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 854 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 855 // CHECK3-NEXT: store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4 856 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 857 // CHECK3: omp.inner.for.cond: 858 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 859 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 860 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 861 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 862 // CHECK3: omp.inner.for.body: 863 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 864 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP15]], 1 865 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 866 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I3]], align 4 867 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[I3]], align 4 868 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* [[TMP1]], i32 0, i32 [[TMP16]] 869 // CHECK3-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 870 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 871 // CHECK3: omp.body.continue: 872 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 873 // CHECK3: omp.inner.for.inc: 874 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 875 // CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP17]], 1 876 // CHECK3-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4 877 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 878 // CHECK3: omp.inner.for.end: 879 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 880 // CHECK3: omp.loop.exit: 881 // CHECK3-NEXT: [[TMP18:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 882 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[TMP18]], align 4 883 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP19]]) 884 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 885 // CHECK3: omp.precond.end: 886 // CHECK3-NEXT: ret void 887 // 888 // 889 // CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 890 // CHECK3-SAME: () #[[ATTR3:[0-9]+]] { 891 // CHECK3-NEXT: entry: 892 // CHECK3-NEXT: call void @__tgt_register_requires(i64 1) 893 // CHECK3-NEXT: ret void 894 // 895 // 896 // CHECK9-LABEL: define {{[^@]+}}@_Z15teams_local_argv 897 // CHECK9-SAME: () #[[ATTR0:[0-9]+]] { 898 // CHECK9-NEXT: entry: 899 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 900 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 901 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 902 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 903 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 904 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 905 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 906 // CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8 907 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 908 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 909 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 910 // CHECK9-NEXT: store i32 100, i32* [[N]], align 4 911 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 912 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 913 // CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() 914 // CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 915 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 916 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 917 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 918 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* 919 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 920 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 921 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 922 // CHECK9-NEXT: [[TMP6:%.*]] = bitcast [3 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 923 // CHECK9-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP6]], i8* align 8 bitcast ([3 x i64]* @.offload_sizes to i8*), i64 24, i1 false) 924 // CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 925 // CHECK9-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64* 926 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP8]], align 8 927 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 928 // CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64* 929 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP10]], align 8 930 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 931 // CHECK9-NEXT: store i8* null, i8** [[TMP11]], align 8 932 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 933 // CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* 934 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 935 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 936 // CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* 937 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 938 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 939 // CHECK9-NEXT: store i8* null, i8** [[TMP16]], align 8 940 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 941 // CHECK9-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** 942 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP18]], align 8 943 // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 944 // CHECK9-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32** 945 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP20]], align 8 946 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 947 // CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP21]], align 8 948 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 949 // CHECK9-NEXT: store i8* null, i8** [[TMP22]], align 8 950 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 951 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 952 // CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 953 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[N]], align 4 954 // CHECK9-NEXT: store i32 [[TMP26]], i32* [[DOTCAPTURE_EXPR_]], align 4 955 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 956 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP27]], 0 957 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 958 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 959 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 960 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 961 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP28]], 1 962 // CHECK9-NEXT: [[TMP29:%.*]] = zext i32 [[ADD]] to i64 963 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP29]]) 964 // CHECK9-NEXT: [[TMP30:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP23]], i8** [[TMP24]], i64* [[TMP25]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) 965 // CHECK9-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0 966 // CHECK9-NEXT: br i1 [[TMP31]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 967 // CHECK9: omp_offload.failed: 968 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] 969 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] 970 // CHECK9: omp_offload.cont: 971 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0 972 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 973 // CHECK9-NEXT: [[TMP33:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 974 // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP33]]) 975 // CHECK9-NEXT: ret i32 [[TMP32]] 976 // 977 // 978 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73 979 // CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 980 // CHECK9-NEXT: entry: 981 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 982 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 983 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 984 // CHECK9-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 985 // CHECK9-NEXT: store i64 [[VLA]], i64* [[VLA_ADDR]], align 8 986 // CHECK9-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 987 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 988 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 989 // CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 990 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[CONV]], i64 [[TMP0]], i32* [[TMP1]]) 991 // CHECK9-NEXT: ret void 992 // 993 // 994 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined. 995 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 996 // CHECK9-NEXT: entry: 997 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 998 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 999 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 1000 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1001 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 1002 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1003 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1004 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1005 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1006 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 1007 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1008 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1009 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1010 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1011 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 1012 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1013 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1014 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 1015 // CHECK9-NEXT: store i64 [[VLA]], i64* [[VLA_ADDR]], align 8 1016 // CHECK9-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 1017 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 1018 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 1019 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 8 1020 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 1021 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 1022 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1023 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1024 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1025 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1026 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1027 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 1028 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1029 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1030 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1031 // CHECK9: omp.precond.then: 1032 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1033 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1034 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_UB]], align 4 1035 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1036 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1037 // CHECK9-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1038 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 1039 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1040 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1041 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1042 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 1043 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1044 // CHECK9: cond.true: 1045 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1046 // CHECK9-NEXT: br label [[COND_END:%.*]] 1047 // CHECK9: cond.false: 1048 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1049 // CHECK9-NEXT: br label [[COND_END]] 1050 // CHECK9: cond.end: 1051 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 1052 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 1053 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1054 // CHECK9-NEXT: store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4 1055 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1056 // CHECK9: omp.inner.for.cond: 1057 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1058 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1059 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 1060 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1061 // CHECK9: omp.inner.for.body: 1062 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1063 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 1064 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1065 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I3]], align 4 1066 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[I3]], align 4 1067 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP17]] to i64 1068 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i64 [[IDXPROM]] 1069 // CHECK9-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 1070 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1071 // CHECK9: omp.body.continue: 1072 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1073 // CHECK9: omp.inner.for.inc: 1074 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1075 // CHECK9-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], 1 1076 // CHECK9-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4 1077 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 1078 // CHECK9: omp.inner.for.end: 1079 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1080 // CHECK9: omp.loop.exit: 1081 // CHECK9-NEXT: [[TMP19:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1082 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4 1083 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP20]]) 1084 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 1085 // CHECK9: omp.precond.end: 1086 // CHECK9-NEXT: ret void 1087 // 1088 // 1089 // CHECK9-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 1090 // CHECK9-SAME: () #[[ATTR5:[0-9]+]] { 1091 // CHECK9-NEXT: entry: 1092 // CHECK9-NEXT: call void @__tgt_register_requires(i64 1) 1093 // CHECK9-NEXT: ret void 1094 // 1095 // 1096 // CHECK11-LABEL: define {{[^@]+}}@_Z15teams_local_argv 1097 // CHECK11-SAME: () #[[ATTR0:[0-9]+]] { 1098 // CHECK11-NEXT: entry: 1099 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 1100 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 1101 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 1102 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 1103 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 1104 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 1105 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 1106 // CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4 1107 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 1108 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1109 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1110 // CHECK11-NEXT: store i32 100, i32* [[N]], align 4 1111 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 1112 // CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() 1113 // CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 1114 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 1115 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 1116 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 1117 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 1118 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 1119 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 1120 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 1121 // CHECK11-NEXT: [[TMP6:%.*]] = bitcast [3 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 1122 // CHECK11-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP6]], i8* align 4 bitcast ([3 x i64]* @.offload_sizes to i8*), i32 24, i1 false) 1123 // CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1124 // CHECK11-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i32* 1125 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP8]], align 4 1126 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1127 // CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32* 1128 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP10]], align 4 1129 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 1130 // CHECK11-NEXT: store i8* null, i8** [[TMP11]], align 4 1131 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 1132 // CHECK11-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* 1133 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 1134 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 1135 // CHECK11-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* 1136 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 1137 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 1138 // CHECK11-NEXT: store i8* null, i8** [[TMP16]], align 4 1139 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 1140 // CHECK11-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** 1141 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP18]], align 4 1142 // CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 1143 // CHECK11-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32** 1144 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP20]], align 4 1145 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 1146 // CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP21]], align 4 1147 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 1148 // CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4 1149 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1150 // CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1151 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 1152 // CHECK11-NEXT: [[TMP26:%.*]] = load i32, i32* [[N]], align 4 1153 // CHECK11-NEXT: store i32 [[TMP26]], i32* [[DOTCAPTURE_EXPR_]], align 4 1154 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1155 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP27]], 0 1156 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1157 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1158 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1159 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1160 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP28]], 1 1161 // CHECK11-NEXT: [[TMP29:%.*]] = zext i32 [[ADD]] to i64 1162 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP29]]) 1163 // CHECK11-NEXT: [[TMP30:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP23]], i8** [[TMP24]], i64* [[TMP25]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) 1164 // CHECK11-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0 1165 // CHECK11-NEXT: br i1 [[TMP31]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1166 // CHECK11: omp_offload.failed: 1167 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] 1168 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] 1169 // CHECK11: omp_offload.cont: 1170 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0 1171 // CHECK11-NEXT: [[TMP32:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 1172 // CHECK11-NEXT: [[TMP33:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 1173 // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP33]]) 1174 // CHECK11-NEXT: ret i32 [[TMP32]] 1175 // 1176 // 1177 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73 1178 // CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 1179 // CHECK11-NEXT: entry: 1180 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 1181 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 1182 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 1183 // CHECK11-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 1184 // CHECK11-NEXT: store i32 [[VLA]], i32* [[VLA_ADDR]], align 4 1185 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 1186 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 1187 // CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 1188 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]]) 1189 // CHECK11-NEXT: ret void 1190 // 1191 // 1192 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined. 1193 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1194 // CHECK11-NEXT: entry: 1195 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 1196 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 1197 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 1198 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 1199 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 1200 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1201 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 1202 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1203 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1204 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 1205 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1206 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1207 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1208 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1209 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 1210 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 1211 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 1212 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 1213 // CHECK11-NEXT: store i32 [[VLA]], i32* [[VLA_ADDR]], align 4 1214 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 1215 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 1216 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 1217 // CHECK11-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 4 1218 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 1219 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 1220 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1221 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1222 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1223 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1224 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1225 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 1226 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1227 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1228 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1229 // CHECK11: omp.precond.then: 1230 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1231 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1232 // CHECK11-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_UB]], align 4 1233 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1234 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1235 // CHECK11-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 1236 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 1237 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1238 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1239 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1240 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 1241 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1242 // CHECK11: cond.true: 1243 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1244 // CHECK11-NEXT: br label [[COND_END:%.*]] 1245 // CHECK11: cond.false: 1246 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1247 // CHECK11-NEXT: br label [[COND_END]] 1248 // CHECK11: cond.end: 1249 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 1250 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 1251 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1252 // CHECK11-NEXT: store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4 1253 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1254 // CHECK11: omp.inner.for.cond: 1255 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1256 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1257 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 1258 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1259 // CHECK11: omp.inner.for.body: 1260 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1261 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 1262 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1263 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I3]], align 4 1264 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[I3]], align 4 1265 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i32 [[TMP17]] 1266 // CHECK11-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 1267 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1268 // CHECK11: omp.body.continue: 1269 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1270 // CHECK11: omp.inner.for.inc: 1271 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1272 // CHECK11-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], 1 1273 // CHECK11-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4 1274 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 1275 // CHECK11: omp.inner.for.end: 1276 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1277 // CHECK11: omp.loop.exit: 1278 // CHECK11-NEXT: [[TMP19:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 1279 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4 1280 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP20]]) 1281 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 1282 // CHECK11: omp.precond.end: 1283 // CHECK11-NEXT: ret void 1284 // 1285 // 1286 // CHECK11-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 1287 // CHECK11-SAME: () #[[ATTR5:[0-9]+]] { 1288 // CHECK11-NEXT: entry: 1289 // CHECK11-NEXT: call void @__tgt_register_requires(i64 1) 1290 // CHECK11-NEXT: ret void 1291 // 1292 // 1293 // CHECK17-LABEL: define {{[^@]+}}@_Z21teams_template_structv 1294 // CHECK17-SAME: () #[[ATTR0:[0-9]+]] { 1295 // CHECK17-NEXT: entry: 1296 // CHECK17-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 1297 // CHECK17-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(%struct.SS* noundef nonnull align 4 dereferenceable(496) [[V]]) 1298 // CHECK17-NEXT: ret i32 [[CALL]] 1299 // 1300 // 1301 // CHECK17-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 1302 // CHECK17-SAME: (%struct.SS* noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 1303 // CHECK17-NEXT: entry: 1304 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 1305 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 1306 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 1307 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 1308 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4 1309 // CHECK17-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 1310 // CHECK17-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 1311 // CHECK17-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0 1312 // CHECK17-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1313 // CHECK17-NEXT: [[TMP1:%.*]] = bitcast i8** [[TMP0]] to %struct.SS** 1314 // CHECK17-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP1]], align 8 1315 // CHECK17-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1316 // CHECK17-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [123 x i32]** 1317 // CHECK17-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP3]], align 8 1318 // CHECK17-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 1319 // CHECK17-NEXT: store i8* null, i8** [[TMP4]], align 8 1320 // CHECK17-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1321 // CHECK17-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1322 // CHECK17-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123) 1323 // CHECK17-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) 1324 // CHECK17-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 1325 // CHECK17-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1326 // CHECK17: omp_offload.failed: 1327 // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]] 1328 // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]] 1329 // CHECK17: omp_offload.cont: 1330 // CHECK17-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 1331 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i64 0, i64 0 1332 // CHECK17-NEXT: [[TMP9:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 1333 // CHECK17-NEXT: ret i32 [[TMP9]] 1334 // 1335 // 1336 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109 1337 // CHECK17-SAME: (%struct.SS* noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 1338 // CHECK17-NEXT: entry: 1339 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 1340 // CHECK17-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 1341 // CHECK17-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 1342 // CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.SS* [[TMP0]]) 1343 // CHECK17-NEXT: ret void 1344 // 1345 // 1346 // CHECK17-LABEL: define {{[^@]+}}@.omp_outlined. 1347 // CHECK17-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %struct.SS* noundef [[THIS:%.*]]) #[[ATTR1]] { 1348 // CHECK17-NEXT: entry: 1349 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1350 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1351 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 1352 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1353 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4 1354 // CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1355 // CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1356 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1357 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1358 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4 1359 // CHECK17-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1360 // CHECK17-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1361 // CHECK17-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 1362 // CHECK17-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 1363 // CHECK17-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1364 // CHECK17-NEXT: store i32 122, i32* [[DOTOMP_UB]], align 4 1365 // CHECK17-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1366 // CHECK17-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1367 // CHECK17-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1368 // CHECK17-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 1369 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1370 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1371 // CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 1372 // CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1373 // CHECK17: cond.true: 1374 // CHECK17-NEXT: br label [[COND_END:%.*]] 1375 // CHECK17: cond.false: 1376 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1377 // CHECK17-NEXT: br label [[COND_END]] 1378 // CHECK17: cond.end: 1379 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 1380 // CHECK17-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 1381 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1382 // CHECK17-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 1383 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1384 // CHECK17: omp.inner.for.cond: 1385 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1386 // CHECK17-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1387 // CHECK17-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 1388 // CHECK17-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1389 // CHECK17: omp.inner.for.body: 1390 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1391 // CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 1392 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1393 // CHECK17-NEXT: store i32 [[ADD]], i32* [[I]], align 4 1394 // CHECK17-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0 1395 // CHECK17-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4 1396 // CHECK17-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP9]] to i64 1397 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A]], i64 0, i64 [[IDXPROM]] 1398 // CHECK17-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 1399 // CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1400 // CHECK17: omp.body.continue: 1401 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1402 // CHECK17: omp.inner.for.inc: 1403 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1404 // CHECK17-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 1405 // CHECK17-NEXT: store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4 1406 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]] 1407 // CHECK17: omp.inner.for.end: 1408 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1409 // CHECK17: omp.loop.exit: 1410 // CHECK17-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) 1411 // CHECK17-NEXT: ret void 1412 // 1413 // 1414 // CHECK17-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 1415 // CHECK17-SAME: () #[[ATTR3:[0-9]+]] { 1416 // CHECK17-NEXT: entry: 1417 // CHECK17-NEXT: call void @__tgt_register_requires(i64 1) 1418 // CHECK17-NEXT: ret void 1419 // 1420 // 1421 // CHECK19-LABEL: define {{[^@]+}}@_Z21teams_template_structv 1422 // CHECK19-SAME: () #[[ATTR0:[0-9]+]] { 1423 // CHECK19-NEXT: entry: 1424 // CHECK19-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 1425 // CHECK19-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(%struct.SS* noundef nonnull align 4 dereferenceable(496) [[V]]) 1426 // CHECK19-NEXT: ret i32 [[CALL]] 1427 // 1428 // 1429 // CHECK19-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 1430 // CHECK19-SAME: (%struct.SS* noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 1431 // CHECK19-NEXT: entry: 1432 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 1433 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 4 1434 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 4 1435 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 4 1436 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4 1437 // CHECK19-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 1438 // CHECK19-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 1439 // CHECK19-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0 1440 // CHECK19-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1441 // CHECK19-NEXT: [[TMP1:%.*]] = bitcast i8** [[TMP0]] to %struct.SS** 1442 // CHECK19-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP1]], align 4 1443 // CHECK19-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1444 // CHECK19-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [123 x i32]** 1445 // CHECK19-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP3]], align 4 1446 // CHECK19-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 1447 // CHECK19-NEXT: store i8* null, i8** [[TMP4]], align 4 1448 // CHECK19-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1449 // CHECK19-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1450 // CHECK19-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 123) 1451 // CHECK19-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) 1452 // CHECK19-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 1453 // CHECK19-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1454 // CHECK19: omp_offload.failed: 1455 // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]] 1456 // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]] 1457 // CHECK19: omp_offload.cont: 1458 // CHECK19-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 1459 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i32 0, i32 0 1460 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 1461 // CHECK19-NEXT: ret i32 [[TMP9]] 1462 // 1463 // 1464 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109 1465 // CHECK19-SAME: (%struct.SS* noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 1466 // CHECK19-NEXT: entry: 1467 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 1468 // CHECK19-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 1469 // CHECK19-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 1470 // CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.SS* [[TMP0]]) 1471 // CHECK19-NEXT: ret void 1472 // 1473 // 1474 // CHECK19-LABEL: define {{[^@]+}}@.omp_outlined. 1475 // CHECK19-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %struct.SS* noundef [[THIS:%.*]]) #[[ATTR1]] { 1476 // CHECK19-NEXT: entry: 1477 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 1478 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 1479 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 1480 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1481 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4 1482 // CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1483 // CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1484 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1485 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1486 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4 1487 // CHECK19-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 1488 // CHECK19-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 1489 // CHECK19-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 1490 // CHECK19-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 1491 // CHECK19-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1492 // CHECK19-NEXT: store i32 122, i32* [[DOTOMP_UB]], align 4 1493 // CHECK19-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1494 // CHECK19-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1495 // CHECK19-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 1496 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 1497 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1498 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1499 // CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 1500 // CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1501 // CHECK19: cond.true: 1502 // CHECK19-NEXT: br label [[COND_END:%.*]] 1503 // CHECK19: cond.false: 1504 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1505 // CHECK19-NEXT: br label [[COND_END]] 1506 // CHECK19: cond.end: 1507 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 1508 // CHECK19-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 1509 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1510 // CHECK19-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 1511 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1512 // CHECK19: omp.inner.for.cond: 1513 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1514 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1515 // CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 1516 // CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1517 // CHECK19: omp.inner.for.body: 1518 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1519 // CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 1520 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1521 // CHECK19-NEXT: store i32 [[ADD]], i32* [[I]], align 4 1522 // CHECK19-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0 1523 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4 1524 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A]], i32 0, i32 [[TMP9]] 1525 // CHECK19-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 1526 // CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1527 // CHECK19: omp.body.continue: 1528 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1529 // CHECK19: omp.inner.for.inc: 1530 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1531 // CHECK19-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 1532 // CHECK19-NEXT: store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4 1533 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]] 1534 // CHECK19: omp.inner.for.end: 1535 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1536 // CHECK19: omp.loop.exit: 1537 // CHECK19-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) 1538 // CHECK19-NEXT: ret void 1539 // 1540 // 1541 // CHECK19-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 1542 // CHECK19-SAME: () #[[ATTR3:[0-9]+]] { 1543 // CHECK19-NEXT: entry: 1544 // CHECK19-NEXT: call void @__tgt_register_requires(i64 1) 1545 // CHECK19-NEXT: ret void 1546 // 1547 // 1548 // CHECK25-LABEL: define {{[^@]+}}@main 1549 // CHECK25-SAME: (i32 noundef signext [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 1550 // CHECK25-NEXT: entry: 1551 // CHECK25-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 1552 // CHECK25-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 1553 // CHECK25-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 1554 // CHECK25-NEXT: [[N:%.*]] = alloca i32, align 4 1555 // CHECK25-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 1556 // CHECK25-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 1557 // CHECK25-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 1558 // CHECK25-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 1559 // CHECK25-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 1560 // CHECK25-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 1561 // CHECK25-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8 1562 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 1563 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1564 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1565 // CHECK25-NEXT: store i32 0, i32* [[RETVAL]], align 4 1566 // CHECK25-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 1567 // CHECK25-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 1568 // CHECK25-NEXT: store i32 100, i32* [[N]], align 4 1569 // CHECK25-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 1570 // CHECK25-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 1571 // CHECK25-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() 1572 // CHECK25-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 1573 // CHECK25-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 1574 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 1575 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 1576 // CHECK25-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* 1577 // CHECK25-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 1578 // CHECK25-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 1579 // CHECK25-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 1580 // CHECK25-NEXT: [[TMP6:%.*]] = bitcast [3 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 1581 // CHECK25-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP6]], i8* align 8 bitcast ([3 x i64]* @.offload_sizes to i8*), i64 24, i1 false) 1582 // CHECK25-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1583 // CHECK25-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64* 1584 // CHECK25-NEXT: store i64 [[TMP4]], i64* [[TMP8]], align 8 1585 // CHECK25-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1586 // CHECK25-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64* 1587 // CHECK25-NEXT: store i64 [[TMP4]], i64* [[TMP10]], align 8 1588 // CHECK25-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 1589 // CHECK25-NEXT: store i8* null, i8** [[TMP11]], align 8 1590 // CHECK25-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 1591 // CHECK25-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* 1592 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 1593 // CHECK25-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 1594 // CHECK25-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* 1595 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 1596 // CHECK25-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 1597 // CHECK25-NEXT: store i8* null, i8** [[TMP16]], align 8 1598 // CHECK25-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 1599 // CHECK25-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** 1600 // CHECK25-NEXT: store i32* [[VLA]], i32** [[TMP18]], align 8 1601 // CHECK25-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 1602 // CHECK25-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32** 1603 // CHECK25-NEXT: store i32* [[VLA]], i32** [[TMP20]], align 8 1604 // CHECK25-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 1605 // CHECK25-NEXT: store i64 [[TMP5]], i64* [[TMP21]], align 8 1606 // CHECK25-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 1607 // CHECK25-NEXT: store i8* null, i8** [[TMP22]], align 8 1608 // CHECK25-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1609 // CHECK25-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1610 // CHECK25-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 1611 // CHECK25-NEXT: [[TMP26:%.*]] = load i32, i32* [[N]], align 4 1612 // CHECK25-NEXT: store i32 [[TMP26]], i32* [[DOTCAPTURE_EXPR_]], align 4 1613 // CHECK25-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1614 // CHECK25-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP27]], 0 1615 // CHECK25-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1616 // CHECK25-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1617 // CHECK25-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1618 // CHECK25-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1619 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP28]], 1 1620 // CHECK25-NEXT: [[TMP29:%.*]] = zext i32 [[ADD]] to i64 1621 // CHECK25-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP29]]) 1622 // CHECK25-NEXT: [[TMP30:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP23]], i8** [[TMP24]], i64* [[TMP25]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) 1623 // CHECK25-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0 1624 // CHECK25-NEXT: br i1 [[TMP31]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1625 // CHECK25: omp_offload.failed: 1626 // CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] 1627 // CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]] 1628 // CHECK25: omp_offload.cont: 1629 // CHECK25-NEXT: [[TMP32:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 1630 // CHECK25-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10EEiT_(i32 noundef signext [[TMP32]]) 1631 // CHECK25-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 1632 // CHECK25-NEXT: [[TMP33:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 1633 // CHECK25-NEXT: call void @llvm.stackrestore(i8* [[TMP33]]) 1634 // CHECK25-NEXT: [[TMP34:%.*]] = load i32, i32* [[RETVAL]], align 4 1635 // CHECK25-NEXT: ret i32 [[TMP34]] 1636 // 1637 // 1638 // CHECK25-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162 1639 // CHECK25-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 1640 // CHECK25-NEXT: entry: 1641 // CHECK25-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 1642 // CHECK25-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1643 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 1644 // CHECK25-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 1645 // CHECK25-NEXT: store i64 [[VLA]], i64* [[VLA_ADDR]], align 8 1646 // CHECK25-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 1647 // CHECK25-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 1648 // CHECK25-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 1649 // CHECK25-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 1650 // CHECK25-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[CONV]], i64 [[TMP0]], i32* [[TMP1]]) 1651 // CHECK25-NEXT: ret void 1652 // 1653 // 1654 // CHECK25-LABEL: define {{[^@]+}}@.omp_outlined. 1655 // CHECK25-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1656 // CHECK25-NEXT: entry: 1657 // CHECK25-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1658 // CHECK25-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1659 // CHECK25-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 1660 // CHECK25-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1661 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 1662 // CHECK25-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1663 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 1664 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1665 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1666 // CHECK25-NEXT: [[I:%.*]] = alloca i32, align 4 1667 // CHECK25-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1668 // CHECK25-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1669 // CHECK25-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1670 // CHECK25-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1671 // CHECK25-NEXT: [[I3:%.*]] = alloca i32, align 4 1672 // CHECK25-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1673 // CHECK25-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1674 // CHECK25-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 1675 // CHECK25-NEXT: store i64 [[VLA]], i64* [[VLA_ADDR]], align 8 1676 // CHECK25-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 1677 // CHECK25-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 1678 // CHECK25-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 1679 // CHECK25-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 8 1680 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 1681 // CHECK25-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 1682 // CHECK25-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1683 // CHECK25-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1684 // CHECK25-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1685 // CHECK25-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1686 // CHECK25-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1687 // CHECK25-NEXT: store i32 0, i32* [[I]], align 4 1688 // CHECK25-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1689 // CHECK25-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1690 // CHECK25-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1691 // CHECK25: omp.precond.then: 1692 // CHECK25-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1693 // CHECK25-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1694 // CHECK25-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_UB]], align 4 1695 // CHECK25-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1696 // CHECK25-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1697 // CHECK25-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1698 // CHECK25-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 1699 // CHECK25-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1700 // CHECK25-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1701 // CHECK25-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1702 // CHECK25-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 1703 // CHECK25-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1704 // CHECK25: cond.true: 1705 // CHECK25-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1706 // CHECK25-NEXT: br label [[COND_END:%.*]] 1707 // CHECK25: cond.false: 1708 // CHECK25-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1709 // CHECK25-NEXT: br label [[COND_END]] 1710 // CHECK25: cond.end: 1711 // CHECK25-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 1712 // CHECK25-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 1713 // CHECK25-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1714 // CHECK25-NEXT: store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4 1715 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1716 // CHECK25: omp.inner.for.cond: 1717 // CHECK25-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1718 // CHECK25-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1719 // CHECK25-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 1720 // CHECK25-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1721 // CHECK25: omp.inner.for.body: 1722 // CHECK25-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1723 // CHECK25-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 1724 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1725 // CHECK25-NEXT: store i32 [[ADD]], i32* [[I3]], align 4 1726 // CHECK25-NEXT: [[TMP17:%.*]] = load i32, i32* [[I3]], align 4 1727 // CHECK25-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP17]] to i64 1728 // CHECK25-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i64 [[IDXPROM]] 1729 // CHECK25-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 1730 // CHECK25-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1731 // CHECK25: omp.body.continue: 1732 // CHECK25-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1733 // CHECK25: omp.inner.for.inc: 1734 // CHECK25-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1735 // CHECK25-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], 1 1736 // CHECK25-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4 1737 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND]] 1738 // CHECK25: omp.inner.for.end: 1739 // CHECK25-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1740 // CHECK25: omp.loop.exit: 1741 // CHECK25-NEXT: [[TMP19:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1742 // CHECK25-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4 1743 // CHECK25-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP20]]) 1744 // CHECK25-NEXT: br label [[OMP_PRECOND_END]] 1745 // CHECK25: omp.precond.end: 1746 // CHECK25-NEXT: ret void 1747 // 1748 // 1749 // CHECK25-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 1750 // CHECK25-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 1751 // CHECK25-NEXT: entry: 1752 // CHECK25-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 1753 // CHECK25-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 1754 // CHECK25-NEXT: [[TE:%.*]] = alloca i32, align 4 1755 // CHECK25-NEXT: [[TH:%.*]] = alloca i32, align 4 1756 // CHECK25-NEXT: [[TE_CASTED:%.*]] = alloca i64, align 8 1757 // CHECK25-NEXT: [[TH_CASTED:%.*]] = alloca i64, align 8 1758 // CHECK25-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 1759 // CHECK25-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 1760 // CHECK25-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 1761 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 1762 // CHECK25-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 1763 // CHECK25-NEXT: store i32 0, i32* [[TE]], align 4 1764 // CHECK25-NEXT: store i32 128, i32* [[TH]], align 4 1765 // CHECK25-NEXT: [[TMP0:%.*]] = load i32, i32* [[TE]], align 4 1766 // CHECK25-NEXT: [[CONV:%.*]] = bitcast i64* [[TE_CASTED]] to i32* 1767 // CHECK25-NEXT: store i32 [[TMP0]], i32* [[CONV]], align 4 1768 // CHECK25-NEXT: [[TMP1:%.*]] = load i64, i64* [[TE_CASTED]], align 8 1769 // CHECK25-NEXT: [[TMP2:%.*]] = load i32, i32* [[TH]], align 4 1770 // CHECK25-NEXT: [[CONV1:%.*]] = bitcast i64* [[TH_CASTED]] to i32* 1771 // CHECK25-NEXT: store i32 [[TMP2]], i32* [[CONV1]], align 4 1772 // CHECK25-NEXT: [[TMP3:%.*]] = load i64, i64* [[TH_CASTED]], align 8 1773 // CHECK25-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1774 // CHECK25-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i64* 1775 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP5]], align 8 1776 // CHECK25-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1777 // CHECK25-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* 1778 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP7]], align 8 1779 // CHECK25-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 1780 // CHECK25-NEXT: store i8* null, i8** [[TMP8]], align 8 1781 // CHECK25-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 1782 // CHECK25-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64* 1783 // CHECK25-NEXT: store i64 [[TMP3]], i64* [[TMP10]], align 8 1784 // CHECK25-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 1785 // CHECK25-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64* 1786 // CHECK25-NEXT: store i64 [[TMP3]], i64* [[TMP12]], align 8 1787 // CHECK25-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 1788 // CHECK25-NEXT: store i8* null, i8** [[TMP13]], align 8 1789 // CHECK25-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 1790 // CHECK25-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to [10 x i32]** 1791 // CHECK25-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP15]], align 8 1792 // CHECK25-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 1793 // CHECK25-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** 1794 // CHECK25-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8 1795 // CHECK25-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 1796 // CHECK25-NEXT: store i8* null, i8** [[TMP18]], align 8 1797 // CHECK25-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1798 // CHECK25-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1799 // CHECK25-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 1800 // CHECK25-NEXT: [[TMP22:%.*]] = load i32, i32* [[TH]], align 4 1801 // CHECK25-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) 1802 // CHECK25-NEXT: [[TMP23:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 [[TMP22]]) 1803 // CHECK25-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 1804 // CHECK25-NEXT: br i1 [[TMP24]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1805 // CHECK25: omp_offload.failed: 1806 // CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] 1807 // CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]] 1808 // CHECK25: omp_offload.cont: 1809 // CHECK25-NEXT: ret i32 0 1810 // 1811 // 1812 // CHECK25-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151 1813 // CHECK25-SAME: (i64 noundef [[TE:%.*]], i64 noundef [[TH:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 1814 // CHECK25-NEXT: entry: 1815 // CHECK25-NEXT: [[TE_ADDR:%.*]] = alloca i64, align 8 1816 // CHECK25-NEXT: [[TH_ADDR:%.*]] = alloca i64, align 8 1817 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 1818 // CHECK25-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) 1819 // CHECK25-NEXT: store i64 [[TE]], i64* [[TE_ADDR]], align 8 1820 // CHECK25-NEXT: store i64 [[TH]], i64* [[TH_ADDR]], align 8 1821 // CHECK25-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 1822 // CHECK25-NEXT: [[CONV:%.*]] = bitcast i64* [[TE_ADDR]] to i32* 1823 // CHECK25-NEXT: [[CONV1:%.*]] = bitcast i64* [[TH_ADDR]] to i32* 1824 // CHECK25-NEXT: [[TMP1:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 1825 // CHECK25-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 1826 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, i32* [[CONV1]], align 4 1827 // CHECK25-NEXT: call void @__kmpc_push_num_teams(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 1828 // CHECK25-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP1]]) 1829 // CHECK25-NEXT: ret void 1830 // 1831 // 1832 // CHECK25-LABEL: define {{[^@]+}}@.omp_outlined..1 1833 // CHECK25-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 1834 // CHECK25-NEXT: entry: 1835 // CHECK25-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1836 // CHECK25-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1837 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 1838 // CHECK25-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1839 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 1840 // CHECK25-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1841 // CHECK25-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1842 // CHECK25-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1843 // CHECK25-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1844 // CHECK25-NEXT: [[I:%.*]] = alloca i32, align 4 1845 // CHECK25-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1846 // CHECK25-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1847 // CHECK25-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 1848 // CHECK25-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 1849 // CHECK25-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1850 // CHECK25-NEXT: store i32 9, i32* [[DOTOMP_UB]], align 4 1851 // CHECK25-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1852 // CHECK25-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1853 // CHECK25-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1854 // CHECK25-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 1855 // CHECK25-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1856 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1857 // CHECK25-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 1858 // CHECK25-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1859 // CHECK25: cond.true: 1860 // CHECK25-NEXT: br label [[COND_END:%.*]] 1861 // CHECK25: cond.false: 1862 // CHECK25-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1863 // CHECK25-NEXT: br label [[COND_END]] 1864 // CHECK25: cond.end: 1865 // CHECK25-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 1866 // CHECK25-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 1867 // CHECK25-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1868 // CHECK25-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 1869 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1870 // CHECK25: omp.inner.for.cond: 1871 // CHECK25-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1872 // CHECK25-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1873 // CHECK25-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 1874 // CHECK25-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1875 // CHECK25: omp.inner.for.body: 1876 // CHECK25-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1877 // CHECK25-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 1878 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1879 // CHECK25-NEXT: store i32 [[ADD]], i32* [[I]], align 4 1880 // CHECK25-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4 1881 // CHECK25-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP9]] to i64 1882 // CHECK25-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], [10 x i32]* [[TMP0]], i64 0, i64 [[IDXPROM]] 1883 // CHECK25-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 1884 // CHECK25-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1885 // CHECK25: omp.body.continue: 1886 // CHECK25-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1887 // CHECK25: omp.inner.for.inc: 1888 // CHECK25-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1889 // CHECK25-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 1890 // CHECK25-NEXT: store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4 1891 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND]] 1892 // CHECK25: omp.inner.for.end: 1893 // CHECK25-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1894 // CHECK25: omp.loop.exit: 1895 // CHECK25-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) 1896 // CHECK25-NEXT: ret void 1897 // 1898 // 1899 // CHECK25-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 1900 // CHECK25-SAME: () #[[ATTR6:[0-9]+]] { 1901 // CHECK25-NEXT: entry: 1902 // CHECK25-NEXT: call void @__tgt_register_requires(i64 1) 1903 // CHECK25-NEXT: ret void 1904 // 1905 // 1906 // CHECK27-LABEL: define {{[^@]+}}@main 1907 // CHECK27-SAME: (i32 noundef [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 1908 // CHECK27-NEXT: entry: 1909 // CHECK27-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 1910 // CHECK27-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 1911 // CHECK27-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 1912 // CHECK27-NEXT: [[N:%.*]] = alloca i32, align 4 1913 // CHECK27-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 1914 // CHECK27-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 1915 // CHECK27-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 1916 // CHECK27-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 1917 // CHECK27-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 1918 // CHECK27-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 1919 // CHECK27-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4 1920 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 1921 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1922 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1923 // CHECK27-NEXT: store i32 0, i32* [[RETVAL]], align 4 1924 // CHECK27-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 1925 // CHECK27-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 1926 // CHECK27-NEXT: store i32 100, i32* [[N]], align 4 1927 // CHECK27-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 1928 // CHECK27-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() 1929 // CHECK27-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 1930 // CHECK27-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 1931 // CHECK27-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 1932 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 1933 // CHECK27-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 1934 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 1935 // CHECK27-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 1936 // CHECK27-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 1937 // CHECK27-NEXT: [[TMP6:%.*]] = bitcast [3 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 1938 // CHECK27-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP6]], i8* align 4 bitcast ([3 x i64]* @.offload_sizes to i8*), i32 24, i1 false) 1939 // CHECK27-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1940 // CHECK27-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i32* 1941 // CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP8]], align 4 1942 // CHECK27-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1943 // CHECK27-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32* 1944 // CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP10]], align 4 1945 // CHECK27-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 1946 // CHECK27-NEXT: store i8* null, i8** [[TMP11]], align 4 1947 // CHECK27-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 1948 // CHECK27-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* 1949 // CHECK27-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 1950 // CHECK27-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 1951 // CHECK27-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* 1952 // CHECK27-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 1953 // CHECK27-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 1954 // CHECK27-NEXT: store i8* null, i8** [[TMP16]], align 4 1955 // CHECK27-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 1956 // CHECK27-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** 1957 // CHECK27-NEXT: store i32* [[VLA]], i32** [[TMP18]], align 4 1958 // CHECK27-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 1959 // CHECK27-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32** 1960 // CHECK27-NEXT: store i32* [[VLA]], i32** [[TMP20]], align 4 1961 // CHECK27-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 1962 // CHECK27-NEXT: store i64 [[TMP5]], i64* [[TMP21]], align 4 1963 // CHECK27-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 1964 // CHECK27-NEXT: store i8* null, i8** [[TMP22]], align 4 1965 // CHECK27-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1966 // CHECK27-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1967 // CHECK27-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 1968 // CHECK27-NEXT: [[TMP26:%.*]] = load i32, i32* [[N]], align 4 1969 // CHECK27-NEXT: store i32 [[TMP26]], i32* [[DOTCAPTURE_EXPR_]], align 4 1970 // CHECK27-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1971 // CHECK27-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP27]], 0 1972 // CHECK27-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1973 // CHECK27-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1974 // CHECK27-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1975 // CHECK27-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1976 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP28]], 1 1977 // CHECK27-NEXT: [[TMP29:%.*]] = zext i32 [[ADD]] to i64 1978 // CHECK27-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2:[0-9]+]], i64 -1, i64 [[TMP29]]) 1979 // CHECK27-NEXT: [[TMP30:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP23]], i8** [[TMP24]], i64* [[TMP25]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) 1980 // CHECK27-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0 1981 // CHECK27-NEXT: br i1 [[TMP31]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1982 // CHECK27: omp_offload.failed: 1983 // CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] 1984 // CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]] 1985 // CHECK27: omp_offload.cont: 1986 // CHECK27-NEXT: [[TMP32:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 1987 // CHECK27-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10EEiT_(i32 noundef [[TMP32]]) 1988 // CHECK27-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 1989 // CHECK27-NEXT: [[TMP33:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 1990 // CHECK27-NEXT: call void @llvm.stackrestore(i8* [[TMP33]]) 1991 // CHECK27-NEXT: [[TMP34:%.*]] = load i32, i32* [[RETVAL]], align 4 1992 // CHECK27-NEXT: ret i32 [[TMP34]] 1993 // 1994 // 1995 // CHECK27-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162 1996 // CHECK27-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 1997 // CHECK27-NEXT: entry: 1998 // CHECK27-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 1999 // CHECK27-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 2000 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 2001 // CHECK27-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 2002 // CHECK27-NEXT: store i32 [[VLA]], i32* [[VLA_ADDR]], align 4 2003 // CHECK27-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 2004 // CHECK27-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 2005 // CHECK27-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 2006 // CHECK27-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]]) 2007 // CHECK27-NEXT: ret void 2008 // 2009 // 2010 // CHECK27-LABEL: define {{[^@]+}}@.omp_outlined. 2011 // CHECK27-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 2012 // CHECK27-NEXT: entry: 2013 // CHECK27-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 2014 // CHECK27-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 2015 // CHECK27-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 2016 // CHECK27-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 2017 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 2018 // CHECK27-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2019 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 2020 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2021 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2022 // CHECK27-NEXT: [[I:%.*]] = alloca i32, align 4 2023 // CHECK27-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2024 // CHECK27-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2025 // CHECK27-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2026 // CHECK27-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2027 // CHECK27-NEXT: [[I3:%.*]] = alloca i32, align 4 2028 // CHECK27-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 2029 // CHECK27-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 2030 // CHECK27-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 2031 // CHECK27-NEXT: store i32 [[VLA]], i32* [[VLA_ADDR]], align 4 2032 // CHECK27-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 2033 // CHECK27-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 2034 // CHECK27-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 2035 // CHECK27-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 4 2036 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 2037 // CHECK27-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 2038 // CHECK27-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2039 // CHECK27-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 2040 // CHECK27-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2041 // CHECK27-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 2042 // CHECK27-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 2043 // CHECK27-NEXT: store i32 0, i32* [[I]], align 4 2044 // CHECK27-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2045 // CHECK27-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 2046 // CHECK27-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2047 // CHECK27: omp.precond.then: 2048 // CHECK27-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 2049 // CHECK27-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2050 // CHECK27-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_UB]], align 4 2051 // CHECK27-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2052 // CHECK27-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2053 // CHECK27-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 2054 // CHECK27-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 2055 // CHECK27-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 2056 // CHECK27-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2057 // CHECK27-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2058 // CHECK27-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 2059 // CHECK27-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2060 // CHECK27: cond.true: 2061 // CHECK27-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2062 // CHECK27-NEXT: br label [[COND_END:%.*]] 2063 // CHECK27: cond.false: 2064 // CHECK27-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2065 // CHECK27-NEXT: br label [[COND_END]] 2066 // CHECK27: cond.end: 2067 // CHECK27-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 2068 // CHECK27-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 2069 // CHECK27-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2070 // CHECK27-NEXT: store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4 2071 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2072 // CHECK27: omp.inner.for.cond: 2073 // CHECK27-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2074 // CHECK27-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2075 // CHECK27-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 2076 // CHECK27-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2077 // CHECK27: omp.inner.for.body: 2078 // CHECK27-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2079 // CHECK27-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 2080 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2081 // CHECK27-NEXT: store i32 [[ADD]], i32* [[I3]], align 4 2082 // CHECK27-NEXT: [[TMP17:%.*]] = load i32, i32* [[I3]], align 4 2083 // CHECK27-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i32 [[TMP17]] 2084 // CHECK27-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 2085 // CHECK27-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2086 // CHECK27: omp.body.continue: 2087 // CHECK27-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2088 // CHECK27: omp.inner.for.inc: 2089 // CHECK27-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2090 // CHECK27-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], 1 2091 // CHECK27-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4 2092 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND]] 2093 // CHECK27: omp.inner.for.end: 2094 // CHECK27-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2095 // CHECK27: omp.loop.exit: 2096 // CHECK27-NEXT: [[TMP19:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 2097 // CHECK27-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4 2098 // CHECK27-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP20]]) 2099 // CHECK27-NEXT: br label [[OMP_PRECOND_END]] 2100 // CHECK27: omp.precond.end: 2101 // CHECK27-NEXT: ret void 2102 // 2103 // 2104 // CHECK27-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 2105 // CHECK27-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 2106 // CHECK27-NEXT: entry: 2107 // CHECK27-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 2108 // CHECK27-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 2109 // CHECK27-NEXT: [[TE:%.*]] = alloca i32, align 4 2110 // CHECK27-NEXT: [[TH:%.*]] = alloca i32, align 4 2111 // CHECK27-NEXT: [[TE_CASTED:%.*]] = alloca i32, align 4 2112 // CHECK27-NEXT: [[TH_CASTED:%.*]] = alloca i32, align 4 2113 // CHECK27-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 2114 // CHECK27-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 2115 // CHECK27-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 2116 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 2117 // CHECK27-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 2118 // CHECK27-NEXT: store i32 0, i32* [[TE]], align 4 2119 // CHECK27-NEXT: store i32 128, i32* [[TH]], align 4 2120 // CHECK27-NEXT: [[TMP0:%.*]] = load i32, i32* [[TE]], align 4 2121 // CHECK27-NEXT: store i32 [[TMP0]], i32* [[TE_CASTED]], align 4 2122 // CHECK27-NEXT: [[TMP1:%.*]] = load i32, i32* [[TE_CASTED]], align 4 2123 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, i32* [[TH]], align 4 2124 // CHECK27-NEXT: store i32 [[TMP2]], i32* [[TH_CASTED]], align 4 2125 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, i32* [[TH_CASTED]], align 4 2126 // CHECK27-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2127 // CHECK27-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i32* 2128 // CHECK27-NEXT: store i32 [[TMP1]], i32* [[TMP5]], align 4 2129 // CHECK27-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2130 // CHECK27-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* 2131 // CHECK27-NEXT: store i32 [[TMP1]], i32* [[TMP7]], align 4 2132 // CHECK27-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 2133 // CHECK27-NEXT: store i8* null, i8** [[TMP8]], align 4 2134 // CHECK27-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 2135 // CHECK27-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32* 2136 // CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP10]], align 4 2137 // CHECK27-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 2138 // CHECK27-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* 2139 // CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP12]], align 4 2140 // CHECK27-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 2141 // CHECK27-NEXT: store i8* null, i8** [[TMP13]], align 4 2142 // CHECK27-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 2143 // CHECK27-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to [10 x i32]** 2144 // CHECK27-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP15]], align 4 2145 // CHECK27-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 2146 // CHECK27-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** 2147 // CHECK27-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4 2148 // CHECK27-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 2149 // CHECK27-NEXT: store i8* null, i8** [[TMP18]], align 4 2150 // CHECK27-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2151 // CHECK27-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2152 // CHECK27-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 2153 // CHECK27-NEXT: [[TMP22:%.*]] = load i32, i32* [[TH]], align 4 2154 // CHECK27-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i64 10) 2155 // CHECK27-NEXT: [[TMP23:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB2]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.2, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.3, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 [[TMP22]]) 2156 // CHECK27-NEXT: [[TMP24:%.*]] = icmp ne i32 [[TMP23]], 0 2157 // CHECK27-NEXT: br i1 [[TMP24]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2158 // CHECK27: omp_offload.failed: 2159 // CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] 2160 // CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]] 2161 // CHECK27: omp_offload.cont: 2162 // CHECK27-NEXT: ret i32 0 2163 // 2164 // 2165 // CHECK27-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151 2166 // CHECK27-SAME: (i32 noundef [[TE:%.*]], i32 noundef [[TH:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2167 // CHECK27-NEXT: entry: 2168 // CHECK27-NEXT: [[TE_ADDR:%.*]] = alloca i32, align 4 2169 // CHECK27-NEXT: [[TH_ADDR:%.*]] = alloca i32, align 4 2170 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 2171 // CHECK27-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB2]]) 2172 // CHECK27-NEXT: store i32 [[TE]], i32* [[TE_ADDR]], align 4 2173 // CHECK27-NEXT: store i32 [[TH]], i32* [[TH_ADDR]], align 4 2174 // CHECK27-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 2175 // CHECK27-NEXT: [[TMP1:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 2176 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, i32* [[TE_ADDR]], align 4 2177 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, i32* [[TH_ADDR]], align 4 2178 // CHECK27-NEXT: call void @__kmpc_push_num_teams(%struct.ident_t* @[[GLOB2]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 2179 // CHECK27-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB2]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP1]]) 2180 // CHECK27-NEXT: ret void 2181 // 2182 // 2183 // CHECK27-LABEL: define {{[^@]+}}@.omp_outlined..1 2184 // CHECK27-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2185 // CHECK27-NEXT: entry: 2186 // CHECK27-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 2187 // CHECK27-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 2188 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 2189 // CHECK27-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2190 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 2191 // CHECK27-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2192 // CHECK27-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2193 // CHECK27-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2194 // CHECK27-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2195 // CHECK27-NEXT: [[I:%.*]] = alloca i32, align 4 2196 // CHECK27-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 2197 // CHECK27-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 2198 // CHECK27-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 2199 // CHECK27-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 2200 // CHECK27-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 2201 // CHECK27-NEXT: store i32 9, i32* [[DOTOMP_UB]], align 4 2202 // CHECK27-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2203 // CHECK27-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2204 // CHECK27-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 2205 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 2206 // CHECK27-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 2207 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2208 // CHECK27-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 2209 // CHECK27-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2210 // CHECK27: cond.true: 2211 // CHECK27-NEXT: br label [[COND_END:%.*]] 2212 // CHECK27: cond.false: 2213 // CHECK27-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2214 // CHECK27-NEXT: br label [[COND_END]] 2215 // CHECK27: cond.end: 2216 // CHECK27-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2217 // CHECK27-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 2218 // CHECK27-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2219 // CHECK27-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 2220 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2221 // CHECK27: omp.inner.for.cond: 2222 // CHECK27-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2223 // CHECK27-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2224 // CHECK27-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 2225 // CHECK27-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2226 // CHECK27: omp.inner.for.body: 2227 // CHECK27-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2228 // CHECK27-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 2229 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2230 // CHECK27-NEXT: store i32 [[ADD]], i32* [[I]], align 4 2231 // CHECK27-NEXT: [[TMP9:%.*]] = load i32, i32* [[I]], align 4 2232 // CHECK27-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], [10 x i32]* [[TMP0]], i32 0, i32 [[TMP9]] 2233 // CHECK27-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 2234 // CHECK27-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2235 // CHECK27: omp.body.continue: 2236 // CHECK27-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2237 // CHECK27: omp.inner.for.inc: 2238 // CHECK27-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2239 // CHECK27-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 2240 // CHECK27-NEXT: store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4 2241 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND]] 2242 // CHECK27: omp.inner.for.end: 2243 // CHECK27-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2244 // CHECK27: omp.loop.exit: 2245 // CHECK27-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) 2246 // CHECK27-NEXT: ret void 2247 // 2248 // 2249 // CHECK27-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 2250 // CHECK27-SAME: () #[[ATTR6:[0-9]+]] { 2251 // CHECK27-NEXT: entry: 2252 // CHECK27-NEXT: call void @__tgt_register_requires(i64 1) 2253 // CHECK27-NEXT: ret void 2254 // 2255