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 6 // Test host codegen. 7 // RUN: %clang_cc1 -DCK1 -verify -Wno-vla -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1 8 // RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 9 // RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1 10 // RUN: %clang_cc1 -DCK1 -verify -Wno-vla -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3 11 // RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 12 // RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK3 13 14 // RUN: %clang_cc1 -DCK1 -verify -Wno-vla -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}}" 15 // RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 16 // RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 17 // RUN: %clang_cc1 -DCK1 -verify -Wno-vla -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}}" 18 // RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 19 // RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 20 #ifdef CK1 21 22 template <typename T, int X, long long Y> 23 struct SS{ 24 T a[X]; 25 float b; 26 int foo(void) { 27 28 #pragma omp target teams distribute parallel for 29 for(int i = 0; i < X; i++) { 30 a[i] = (T)0; 31 } 32 #pragma omp target teams distribute parallel for dist_schedule(static) 33 for(int i = 0; i < X; i++) { 34 a[i] = (T)0; 35 } 36 #pragma omp target teams distribute parallel for dist_schedule(static, X/2) 37 for(int i = 0; i < X; i++) { 38 a[i] = (T)0; 39 } 40 41 42 43 44 45 46 47 48 49 50 return a[0]; 51 } 52 }; 53 54 int teams_template_struct(void) { 55 SS<int, 123, 456> V; 56 return V.foo(); 57 58 } 59 #endif // CK1 60 61 // Test host codegen. 62 // RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK9 63 // RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 64 // RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK9 65 // RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK11 66 // RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 67 // RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK11 68 69 // RUN: %clang_cc1 -DCK2 -verify -Wno-vla -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}}" 70 // RUN: %clang_cc1 -DCK2 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 71 // RUN: %clang_cc1 -DCK2 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 72 // RUN: %clang_cc1 -DCK2 -verify -Wno-vla -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}}" 73 // RUN: %clang_cc1 -DCK2 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 74 // RUN: %clang_cc1 -DCK2 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 75 #ifdef CK2 76 77 template <typename T, int n> 78 int tmain(T argc) { 79 T a[n]; 80 int m = 10; 81 #pragma omp target teams distribute parallel for 82 for(int i = 0; i < n; i++) { 83 a[i] = (T)0; 84 } 85 #pragma omp target teams distribute parallel for dist_schedule(static) 86 for(int i = 0; i < n; i++) { 87 a[i] = (T)0; 88 } 89 #pragma omp target teams distribute parallel for dist_schedule(static, m) 90 for(int i = 0; i < n; i++) { 91 a[i] = (T)0; 92 } 93 return 0; 94 } 95 96 int main (int argc, char **argv) { 97 int n = 100; 98 int a[n]; 99 int m = 10; 100 #pragma omp target teams distribute parallel for 101 for(int i = 0; i < n; i++) { 102 a[i] = 0; 103 } 104 #pragma omp target teams distribute parallel for dist_schedule(static) 105 for(int i = 0; i < n; i++) { 106 a[i] = 0; 107 } 108 #pragma omp target teams distribute parallel for dist_schedule(static, m) 109 for(int i = 0; i < n; i++) { 110 a[i] = 0; 111 } 112 return tmain<int, 10>(argc); 113 } 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 #endif // CK2 136 #endif // #ifndef HEADER 137 // CHECK1-LABEL: define {{[^@]+}}@_Z21teams_template_structv 138 // CHECK1-SAME: () #[[ATTR0:[0-9]+]] { 139 // CHECK1-NEXT: entry: 140 // CHECK1-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 141 // CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]]) 142 // CHECK1-NEXT: ret i32 [[CALL]] 143 // 144 // 145 // CHECK1-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 146 // CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat { 147 // CHECK1-NEXT: entry: 148 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 149 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 150 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 151 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 152 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 153 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 154 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x ptr], align 8 155 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x ptr], align 8 156 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x ptr], align 8 157 // CHECK1-NEXT: [[_TMP6:%.*]] = alloca i32, align 4 158 // CHECK1-NEXT: [[KERNEL_ARGS7:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 159 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS11:%.*]] = alloca [1 x ptr], align 8 160 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS12:%.*]] = alloca [1 x ptr], align 8 161 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS13:%.*]] = alloca [1 x ptr], align 8 162 // CHECK1-NEXT: [[_TMP14:%.*]] = alloca i32, align 4 163 // CHECK1-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 164 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 165 // CHECK1-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 166 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 167 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 168 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 8 169 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 170 // CHECK1-NEXT: store ptr [[A]], ptr [[TMP1]], align 8 171 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 172 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8 173 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 174 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 175 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 176 // CHECK1-NEXT: store i32 3, ptr [[TMP5]], align 4 177 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 178 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4 179 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 180 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8 181 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 182 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8 183 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 184 // CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 8 185 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 186 // CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 8 187 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 188 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8 189 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 190 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8 191 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 192 // CHECK1-NEXT: store i64 123, ptr [[TMP13]], align 8 193 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 194 // CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8 195 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 196 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4 197 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 198 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4 199 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 200 // CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4 201 // CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]]) 202 // CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0 203 // CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 204 // CHECK1: omp_offload.failed: 205 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]] 206 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] 207 // CHECK1: omp_offload.cont: 208 // CHECK1-NEXT: [[A2:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 209 // CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 210 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP20]], align 8 211 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 212 // CHECK1-NEXT: store ptr [[A2]], ptr [[TMP21]], align 8 213 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 0 214 // CHECK1-NEXT: store ptr null, ptr [[TMP22]], align 8 215 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 216 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 217 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 0 218 // CHECK1-NEXT: store i32 3, ptr [[TMP25]], align 4 219 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 1 220 // CHECK1-NEXT: store i32 1, ptr [[TMP26]], align 4 221 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 2 222 // CHECK1-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 8 223 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 3 224 // CHECK1-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 8 225 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 4 226 // CHECK1-NEXT: store ptr @.offload_sizes.1, ptr [[TMP29]], align 8 227 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 5 228 // CHECK1-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP30]], align 8 229 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 6 230 // CHECK1-NEXT: store ptr null, ptr [[TMP31]], align 8 231 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 7 232 // CHECK1-NEXT: store ptr null, ptr [[TMP32]], align 8 233 // CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 8 234 // CHECK1-NEXT: store i64 123, ptr [[TMP33]], align 8 235 // CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 9 236 // CHECK1-NEXT: store i64 0, ptr [[TMP34]], align 8 237 // CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 10 238 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4 239 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 11 240 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4 241 // CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 12 242 // CHECK1-NEXT: store i32 0, ptr [[TMP37]], align 4 243 // CHECK1-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l32.region_id, ptr [[KERNEL_ARGS7]]) 244 // CHECK1-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0 245 // CHECK1-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED8:%.*]], label [[OMP_OFFLOAD_CONT9:%.*]] 246 // CHECK1: omp_offload.failed8: 247 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l32(ptr [[THIS1]]) #[[ATTR2]] 248 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT9]] 249 // CHECK1: omp_offload.cont9: 250 // CHECK1-NEXT: [[A10:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 251 // CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0 252 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP40]], align 8 253 // CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0 254 // CHECK1-NEXT: store ptr [[A10]], ptr [[TMP41]], align 8 255 // CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS13]], i64 0, i64 0 256 // CHECK1-NEXT: store ptr null, ptr [[TMP42]], align 8 257 // CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0 258 // CHECK1-NEXT: [[TMP44:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0 259 // CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0 260 // CHECK1-NEXT: store i32 3, ptr [[TMP45]], align 4 261 // CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1 262 // CHECK1-NEXT: store i32 1, ptr [[TMP46]], align 4 263 // CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2 264 // CHECK1-NEXT: store ptr [[TMP43]], ptr [[TMP47]], align 8 265 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3 266 // CHECK1-NEXT: store ptr [[TMP44]], ptr [[TMP48]], align 8 267 // CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4 268 // CHECK1-NEXT: store ptr @.offload_sizes.3, ptr [[TMP49]], align 8 269 // CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5 270 // CHECK1-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP50]], align 8 271 // CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6 272 // CHECK1-NEXT: store ptr null, ptr [[TMP51]], align 8 273 // CHECK1-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7 274 // CHECK1-NEXT: store ptr null, ptr [[TMP52]], align 8 275 // CHECK1-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8 276 // CHECK1-NEXT: store i64 123, ptr [[TMP53]], align 8 277 // CHECK1-NEXT: [[TMP54:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9 278 // CHECK1-NEXT: store i64 0, ptr [[TMP54]], align 8 279 // CHECK1-NEXT: [[TMP55:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10 280 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP55]], align 4 281 // CHECK1-NEXT: [[TMP56:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11 282 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP56]], align 4 283 // CHECK1-NEXT: [[TMP57:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12 284 // CHECK1-NEXT: store i32 0, ptr [[TMP57]], align 4 285 // CHECK1-NEXT: [[TMP58:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l36.region_id, ptr [[KERNEL_ARGS15]]) 286 // CHECK1-NEXT: [[TMP59:%.*]] = icmp ne i32 [[TMP58]], 0 287 // CHECK1-NEXT: br i1 [[TMP59]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] 288 // CHECK1: omp_offload.failed16: 289 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l36(ptr [[THIS1]]) #[[ATTR2]] 290 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT17]] 291 // CHECK1: omp_offload.cont17: 292 // CHECK1-NEXT: [[A18:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 293 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A18]], i64 0, i64 0 294 // CHECK1-NEXT: [[TMP60:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 295 // CHECK1-NEXT: ret i32 [[TMP60]] 296 // 297 // 298 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28 299 // CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 300 // CHECK1-NEXT: entry: 301 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 302 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 303 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 304 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined, ptr [[TMP0]]) 305 // CHECK1-NEXT: ret void 306 // 307 // 308 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined 309 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 310 // CHECK1-NEXT: entry: 311 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 312 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 313 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 314 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 315 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 316 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 317 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 318 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 319 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 320 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 321 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 322 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 323 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 324 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 325 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 326 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4 327 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 328 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 329 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 330 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 331 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 332 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 333 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 334 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 335 // CHECK1: cond.true: 336 // CHECK1-NEXT: br label [[COND_END:%.*]] 337 // CHECK1: cond.false: 338 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 339 // CHECK1-NEXT: br label [[COND_END]] 340 // CHECK1: cond.end: 341 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 342 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 343 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 344 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 345 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 346 // CHECK1: omp.inner.for.cond: 347 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 348 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 349 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 350 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 351 // CHECK1: omp.inner.for.body: 352 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 353 // CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 354 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 355 // CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 356 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]]) 357 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 358 // CHECK1: omp.inner.for.inc: 359 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 360 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 361 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] 362 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 363 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 364 // CHECK1: omp.inner.for.end: 365 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 366 // CHECK1: omp.loop.exit: 367 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 368 // CHECK1-NEXT: ret void 369 // 370 // 371 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined.omp_outlined 372 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 373 // CHECK1-NEXT: entry: 374 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 375 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 376 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 377 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 378 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 379 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 380 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 381 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 382 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 383 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 384 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 385 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 386 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 387 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 388 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8 389 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8 390 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 391 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 392 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 393 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 394 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8 395 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32 396 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 397 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32 398 // CHECK1-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4 399 // CHECK1-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4 400 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 401 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 402 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 403 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 404 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 405 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 406 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122 407 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 408 // CHECK1: cond.true: 409 // CHECK1-NEXT: br label [[COND_END:%.*]] 410 // CHECK1: cond.false: 411 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 412 // CHECK1-NEXT: br label [[COND_END]] 413 // CHECK1: cond.end: 414 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 415 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 416 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 417 // CHECK1-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 418 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 419 // CHECK1: omp.inner.for.cond: 420 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 421 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 422 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 423 // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 424 // CHECK1: omp.inner.for.body: 425 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 426 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 427 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 428 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4 429 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 430 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4 431 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 432 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]] 433 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 434 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 435 // CHECK1: omp.body.continue: 436 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 437 // CHECK1: omp.inner.for.inc: 438 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 439 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1 440 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4 441 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 442 // CHECK1: omp.inner.for.end: 443 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 444 // CHECK1: omp.loop.exit: 445 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]]) 446 // CHECK1-NEXT: ret void 447 // 448 // 449 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l32 450 // CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 451 // CHECK1-NEXT: entry: 452 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 453 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 454 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 455 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l32.omp_outlined, ptr [[TMP0]]) 456 // CHECK1-NEXT: ret void 457 // 458 // 459 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l32.omp_outlined 460 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 461 // CHECK1-NEXT: entry: 462 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 463 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 464 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 465 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 466 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 467 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 468 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 469 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 470 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 471 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 472 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 473 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 474 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 475 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 476 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 477 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4 478 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 479 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 480 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 481 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 482 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 483 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 484 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 485 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 486 // CHECK1: cond.true: 487 // CHECK1-NEXT: br label [[COND_END:%.*]] 488 // CHECK1: cond.false: 489 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 490 // CHECK1-NEXT: br label [[COND_END]] 491 // CHECK1: cond.end: 492 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 493 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 494 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 495 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 496 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 497 // CHECK1: omp.inner.for.cond: 498 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 499 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 500 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 501 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 502 // CHECK1: omp.inner.for.body: 503 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 504 // CHECK1-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 505 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 506 // CHECK1-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 507 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l32.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]]) 508 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 509 // CHECK1: omp.inner.for.inc: 510 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 511 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 512 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] 513 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 514 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 515 // CHECK1: omp.inner.for.end: 516 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 517 // CHECK1: omp.loop.exit: 518 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 519 // CHECK1-NEXT: ret void 520 // 521 // 522 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l32.omp_outlined.omp_outlined 523 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 524 // CHECK1-NEXT: entry: 525 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 526 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 527 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 528 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 529 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 530 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 531 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 532 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 533 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 534 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 535 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 536 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 537 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 538 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 539 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8 540 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8 541 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 542 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 543 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 544 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 545 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8 546 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32 547 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 548 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32 549 // CHECK1-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4 550 // CHECK1-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4 551 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 552 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 553 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 554 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 555 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 556 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 557 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122 558 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 559 // CHECK1: cond.true: 560 // CHECK1-NEXT: br label [[COND_END:%.*]] 561 // CHECK1: cond.false: 562 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 563 // CHECK1-NEXT: br label [[COND_END]] 564 // CHECK1: cond.end: 565 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 566 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 567 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 568 // CHECK1-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 569 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 570 // CHECK1: omp.inner.for.cond: 571 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 572 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 573 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 574 // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 575 // CHECK1: omp.inner.for.body: 576 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 577 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 578 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 579 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4 580 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 581 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4 582 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 583 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]] 584 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 585 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 586 // CHECK1: omp.body.continue: 587 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 588 // CHECK1: omp.inner.for.inc: 589 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 590 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1 591 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4 592 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 593 // CHECK1: omp.inner.for.end: 594 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 595 // CHECK1: omp.loop.exit: 596 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]]) 597 // CHECK1-NEXT: ret void 598 // 599 // 600 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l36 601 // CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 602 // CHECK1-NEXT: entry: 603 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 604 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 605 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 606 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l36.omp_outlined, ptr [[TMP0]]) 607 // CHECK1-NEXT: ret void 608 // 609 // 610 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l36.omp_outlined 611 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 612 // CHECK1-NEXT: entry: 613 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 614 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 615 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 616 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 617 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 618 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 619 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 620 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 621 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 622 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 623 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 624 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 625 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 626 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 627 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 628 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4 629 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 630 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 631 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 632 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 633 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 61) 634 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 635 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 636 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 637 // CHECK1: cond.true: 638 // CHECK1-NEXT: br label [[COND_END:%.*]] 639 // CHECK1: cond.false: 640 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 641 // CHECK1-NEXT: br label [[COND_END]] 642 // CHECK1: cond.end: 643 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 644 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 645 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 646 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 647 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 648 // CHECK1: omp.inner.for.cond: 649 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 650 // CHECK1-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP6]], 123 651 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 652 // CHECK1: omp.inner.for.body: 653 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 654 // CHECK1-NEXT: [[TMP8:%.*]] = zext i32 [[TMP7]] to i64 655 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 656 // CHECK1-NEXT: [[TMP10:%.*]] = zext i32 [[TMP9]] to i64 657 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l36.omp_outlined.omp_outlined, i64 [[TMP8]], i64 [[TMP10]], ptr [[TMP0]]) 658 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 659 // CHECK1: omp.inner.for.inc: 660 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 661 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 662 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] 663 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 664 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 665 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 666 // CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP13]], [[TMP14]] 667 // CHECK1-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4 668 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 669 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 670 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP15]], [[TMP16]] 671 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4 672 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 673 // CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP17]], 122 674 // CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]] 675 // CHECK1: cond.true5: 676 // CHECK1-NEXT: br label [[COND_END7:%.*]] 677 // CHECK1: cond.false6: 678 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 679 // CHECK1-NEXT: br label [[COND_END7]] 680 // CHECK1: cond.end7: 681 // CHECK1-NEXT: [[COND8:%.*]] = phi i32 [ 122, [[COND_TRUE5]] ], [ [[TMP18]], [[COND_FALSE6]] ] 682 // CHECK1-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4 683 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 684 // CHECK1-NEXT: store i32 [[TMP19]], ptr [[DOTOMP_IV]], align 4 685 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 686 // CHECK1: omp.inner.for.end: 687 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 688 // CHECK1: omp.loop.exit: 689 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 690 // CHECK1-NEXT: ret void 691 // 692 // 693 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l36.omp_outlined.omp_outlined 694 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 695 // CHECK1-NEXT: entry: 696 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 697 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 698 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 699 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 700 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 701 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 702 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 703 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 704 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 705 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 706 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 707 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 708 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 709 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 710 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8 711 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8 712 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 713 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 714 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 715 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 716 // CHECK1-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8 717 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32 718 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 719 // CHECK1-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32 720 // CHECK1-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4 721 // CHECK1-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4 722 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 723 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 724 // CHECK1-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 725 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 726 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 727 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 728 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122 729 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 730 // CHECK1: cond.true: 731 // CHECK1-NEXT: br label [[COND_END:%.*]] 732 // CHECK1: cond.false: 733 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 734 // CHECK1-NEXT: br label [[COND_END]] 735 // CHECK1: cond.end: 736 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 737 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 738 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 739 // CHECK1-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 740 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 741 // CHECK1: omp.inner.for.cond: 742 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 743 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 744 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 745 // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 746 // CHECK1: omp.inner.for.body: 747 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 748 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 749 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 750 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4 751 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 752 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4 753 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 754 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]] 755 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 756 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 757 // CHECK1: omp.body.continue: 758 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 759 // CHECK1: omp.inner.for.inc: 760 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 761 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1 762 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4 763 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 764 // CHECK1: omp.inner.for.end: 765 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 766 // CHECK1: omp.loop.exit: 767 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]]) 768 // CHECK1-NEXT: ret void 769 // 770 // 771 // CHECK3-LABEL: define {{[^@]+}}@_Z21teams_template_structv 772 // CHECK3-SAME: () #[[ATTR0:[0-9]+]] { 773 // CHECK3-NEXT: entry: 774 // CHECK3-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 775 // CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]]) 776 // CHECK3-NEXT: ret i32 [[CALL]] 777 // 778 // 779 // CHECK3-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 780 // CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 781 // CHECK3-NEXT: entry: 782 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 783 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 784 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 785 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 786 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 787 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 788 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x ptr], align 4 789 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x ptr], align 4 790 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x ptr], align 4 791 // CHECK3-NEXT: [[_TMP6:%.*]] = alloca i32, align 4 792 // CHECK3-NEXT: [[KERNEL_ARGS7:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 793 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS11:%.*]] = alloca [1 x ptr], align 4 794 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS12:%.*]] = alloca [1 x ptr], align 4 795 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS13:%.*]] = alloca [1 x ptr], align 4 796 // CHECK3-NEXT: [[_TMP14:%.*]] = alloca i32, align 4 797 // CHECK3-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 798 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 799 // CHECK3-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 800 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 801 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 802 // CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 4 803 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 804 // CHECK3-NEXT: store ptr [[A]], ptr [[TMP1]], align 4 805 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 806 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4 807 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 808 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 809 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 810 // CHECK3-NEXT: store i32 3, ptr [[TMP5]], align 4 811 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 812 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4 813 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 814 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4 815 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 816 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4 817 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 818 // CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 4 819 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 820 // CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 4 821 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 822 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4 823 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 824 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4 825 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 826 // CHECK3-NEXT: store i64 123, ptr [[TMP13]], align 8 827 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 828 // CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8 829 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 830 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4 831 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 832 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4 833 // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 834 // CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4 835 // CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]]) 836 // CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0 837 // CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 838 // CHECK3: omp_offload.failed: 839 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]] 840 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] 841 // CHECK3: omp_offload.cont: 842 // CHECK3-NEXT: [[A2:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 843 // CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 844 // CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP20]], align 4 845 // CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 846 // CHECK3-NEXT: store ptr [[A2]], ptr [[TMP21]], align 4 847 // CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS5]], i32 0, i32 0 848 // CHECK3-NEXT: store ptr null, ptr [[TMP22]], align 4 849 // CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 850 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 851 // CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 0 852 // CHECK3-NEXT: store i32 3, ptr [[TMP25]], align 4 853 // CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 1 854 // CHECK3-NEXT: store i32 1, ptr [[TMP26]], align 4 855 // CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 2 856 // CHECK3-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 4 857 // CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 3 858 // CHECK3-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 4 859 // CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 4 860 // CHECK3-NEXT: store ptr @.offload_sizes.1, ptr [[TMP29]], align 4 861 // CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 5 862 // CHECK3-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP30]], align 4 863 // CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 6 864 // CHECK3-NEXT: store ptr null, ptr [[TMP31]], align 4 865 // CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 7 866 // CHECK3-NEXT: store ptr null, ptr [[TMP32]], align 4 867 // CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 8 868 // CHECK3-NEXT: store i64 123, ptr [[TMP33]], align 8 869 // CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 9 870 // CHECK3-NEXT: store i64 0, ptr [[TMP34]], align 8 871 // CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 10 872 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4 873 // CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 11 874 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4 875 // CHECK3-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 12 876 // CHECK3-NEXT: store i32 0, ptr [[TMP37]], align 4 877 // CHECK3-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l32.region_id, ptr [[KERNEL_ARGS7]]) 878 // CHECK3-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0 879 // CHECK3-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED8:%.*]], label [[OMP_OFFLOAD_CONT9:%.*]] 880 // CHECK3: omp_offload.failed8: 881 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l32(ptr [[THIS1]]) #[[ATTR2]] 882 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT9]] 883 // CHECK3: omp_offload.cont9: 884 // CHECK3-NEXT: [[A10:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 885 // CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0 886 // CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP40]], align 4 887 // CHECK3-NEXT: [[TMP41:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0 888 // CHECK3-NEXT: store ptr [[A10]], ptr [[TMP41]], align 4 889 // CHECK3-NEXT: [[TMP42:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS13]], i32 0, i32 0 890 // CHECK3-NEXT: store ptr null, ptr [[TMP42]], align 4 891 // CHECK3-NEXT: [[TMP43:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0 892 // CHECK3-NEXT: [[TMP44:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0 893 // CHECK3-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0 894 // CHECK3-NEXT: store i32 3, ptr [[TMP45]], align 4 895 // CHECK3-NEXT: [[TMP46:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1 896 // CHECK3-NEXT: store i32 1, ptr [[TMP46]], align 4 897 // CHECK3-NEXT: [[TMP47:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2 898 // CHECK3-NEXT: store ptr [[TMP43]], ptr [[TMP47]], align 4 899 // CHECK3-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3 900 // CHECK3-NEXT: store ptr [[TMP44]], ptr [[TMP48]], align 4 901 // CHECK3-NEXT: [[TMP49:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4 902 // CHECK3-NEXT: store ptr @.offload_sizes.3, ptr [[TMP49]], align 4 903 // CHECK3-NEXT: [[TMP50:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5 904 // CHECK3-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP50]], align 4 905 // CHECK3-NEXT: [[TMP51:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6 906 // CHECK3-NEXT: store ptr null, ptr [[TMP51]], align 4 907 // CHECK3-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7 908 // CHECK3-NEXT: store ptr null, ptr [[TMP52]], align 4 909 // CHECK3-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8 910 // CHECK3-NEXT: store i64 123, ptr [[TMP53]], align 8 911 // CHECK3-NEXT: [[TMP54:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9 912 // CHECK3-NEXT: store i64 0, ptr [[TMP54]], align 8 913 // CHECK3-NEXT: [[TMP55:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10 914 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP55]], align 4 915 // CHECK3-NEXT: [[TMP56:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11 916 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP56]], align 4 917 // CHECK3-NEXT: [[TMP57:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12 918 // CHECK3-NEXT: store i32 0, ptr [[TMP57]], align 4 919 // CHECK3-NEXT: [[TMP58:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l36.region_id, ptr [[KERNEL_ARGS15]]) 920 // CHECK3-NEXT: [[TMP59:%.*]] = icmp ne i32 [[TMP58]], 0 921 // CHECK3-NEXT: br i1 [[TMP59]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] 922 // CHECK3: omp_offload.failed16: 923 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l36(ptr [[THIS1]]) #[[ATTR2]] 924 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT17]] 925 // CHECK3: omp_offload.cont17: 926 // CHECK3-NEXT: [[A18:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 927 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A18]], i32 0, i32 0 928 // CHECK3-NEXT: [[TMP60:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 929 // CHECK3-NEXT: ret i32 [[TMP60]] 930 // 931 // 932 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28 933 // CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 934 // CHECK3-NEXT: entry: 935 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 936 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 937 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 938 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined, ptr [[TMP0]]) 939 // CHECK3-NEXT: ret void 940 // 941 // 942 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined 943 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 944 // CHECK3-NEXT: entry: 945 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 946 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 947 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 948 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 949 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 950 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 951 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 952 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 953 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 954 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 955 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 956 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 957 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 958 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 959 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 960 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4 961 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 962 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 963 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 964 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 965 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 966 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 967 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 968 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 969 // CHECK3: cond.true: 970 // CHECK3-NEXT: br label [[COND_END:%.*]] 971 // CHECK3: cond.false: 972 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 973 // CHECK3-NEXT: br label [[COND_END]] 974 // CHECK3: cond.end: 975 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 976 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 977 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 978 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 979 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 980 // CHECK3: omp.inner.for.cond: 981 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 982 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 983 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 984 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 985 // CHECK3: omp.inner.for.body: 986 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 987 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 988 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]]) 989 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 990 // CHECK3: omp.inner.for.inc: 991 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 992 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 993 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] 994 // CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 995 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 996 // CHECK3: omp.inner.for.end: 997 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 998 // CHECK3: omp.loop.exit: 999 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 1000 // CHECK3-NEXT: ret void 1001 // 1002 // 1003 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined.omp_outlined 1004 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 1005 // CHECK3-NEXT: entry: 1006 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 1007 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 1008 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 1009 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 1010 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 1011 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1012 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 1013 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1014 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1015 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1016 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1017 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 1018 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 1019 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 1020 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4 1021 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4 1022 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 1023 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 1024 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1025 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 1026 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4 1027 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4 1028 // CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4 1029 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4 1030 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1031 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1032 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1033 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 1034 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1035 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1036 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122 1037 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1038 // CHECK3: cond.true: 1039 // CHECK3-NEXT: br label [[COND_END:%.*]] 1040 // CHECK3: cond.false: 1041 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1042 // CHECK3-NEXT: br label [[COND_END]] 1043 // CHECK3: cond.end: 1044 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 1045 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 1046 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1047 // CHECK3-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 1048 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1049 // CHECK3: omp.inner.for.cond: 1050 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1051 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1052 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 1053 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1054 // CHECK3: omp.inner.for.body: 1055 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1056 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 1057 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1058 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4 1059 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 1060 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4 1061 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP11]] 1062 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 1063 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1064 // CHECK3: omp.body.continue: 1065 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1066 // CHECK3: omp.inner.for.inc: 1067 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1068 // CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1 1069 // CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 1070 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 1071 // CHECK3: omp.inner.for.end: 1072 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1073 // CHECK3: omp.loop.exit: 1074 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]]) 1075 // CHECK3-NEXT: ret void 1076 // 1077 // 1078 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l32 1079 // CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 1080 // CHECK3-NEXT: entry: 1081 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 1082 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 1083 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 1084 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l32.omp_outlined, ptr [[TMP0]]) 1085 // CHECK3-NEXT: ret void 1086 // 1087 // 1088 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l32.omp_outlined 1089 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 1090 // CHECK3-NEXT: entry: 1091 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 1092 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 1093 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 1094 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1095 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 1096 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 1097 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 1098 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1099 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1100 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 1101 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 1102 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 1103 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 1104 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 1105 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 1106 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4 1107 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1108 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1109 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1110 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 1111 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1112 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1113 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 1114 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1115 // CHECK3: cond.true: 1116 // CHECK3-NEXT: br label [[COND_END:%.*]] 1117 // CHECK3: cond.false: 1118 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1119 // CHECK3-NEXT: br label [[COND_END]] 1120 // CHECK3: cond.end: 1121 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 1122 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 1123 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1124 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 1125 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1126 // CHECK3: omp.inner.for.cond: 1127 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1128 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1129 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 1130 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1131 // CHECK3: omp.inner.for.body: 1132 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1133 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1134 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l32.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]]) 1135 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1136 // CHECK3: omp.inner.for.inc: 1137 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1138 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 1139 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] 1140 // CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 1141 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 1142 // CHECK3: omp.inner.for.end: 1143 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1144 // CHECK3: omp.loop.exit: 1145 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 1146 // CHECK3-NEXT: ret void 1147 // 1148 // 1149 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l32.omp_outlined.omp_outlined 1150 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 1151 // CHECK3-NEXT: entry: 1152 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 1153 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 1154 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 1155 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 1156 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 1157 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1158 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 1159 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1160 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1161 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1162 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1163 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 1164 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 1165 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 1166 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4 1167 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4 1168 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 1169 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 1170 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1171 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 1172 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4 1173 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4 1174 // CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4 1175 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4 1176 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1177 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1178 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1179 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 1180 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1181 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1182 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122 1183 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1184 // CHECK3: cond.true: 1185 // CHECK3-NEXT: br label [[COND_END:%.*]] 1186 // CHECK3: cond.false: 1187 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1188 // CHECK3-NEXT: br label [[COND_END]] 1189 // CHECK3: cond.end: 1190 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 1191 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 1192 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1193 // CHECK3-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 1194 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1195 // CHECK3: omp.inner.for.cond: 1196 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1197 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1198 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 1199 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1200 // CHECK3: omp.inner.for.body: 1201 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1202 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 1203 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1204 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4 1205 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 1206 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4 1207 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP11]] 1208 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 1209 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1210 // CHECK3: omp.body.continue: 1211 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1212 // CHECK3: omp.inner.for.inc: 1213 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1214 // CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1 1215 // CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 1216 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 1217 // CHECK3: omp.inner.for.end: 1218 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1219 // CHECK3: omp.loop.exit: 1220 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]]) 1221 // CHECK3-NEXT: ret void 1222 // 1223 // 1224 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l36 1225 // CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 1226 // CHECK3-NEXT: entry: 1227 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 1228 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 1229 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 1230 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l36.omp_outlined, ptr [[TMP0]]) 1231 // CHECK3-NEXT: ret void 1232 // 1233 // 1234 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l36.omp_outlined 1235 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 1236 // CHECK3-NEXT: entry: 1237 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 1238 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 1239 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 1240 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1241 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 1242 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 1243 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 1244 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1245 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1246 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 1247 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 1248 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 1249 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 1250 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 1251 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 1252 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4 1253 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1254 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1255 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1256 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 1257 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 61) 1258 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1259 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 1260 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1261 // CHECK3: cond.true: 1262 // CHECK3-NEXT: br label [[COND_END:%.*]] 1263 // CHECK3: cond.false: 1264 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1265 // CHECK3-NEXT: br label [[COND_END]] 1266 // CHECK3: cond.end: 1267 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 1268 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 1269 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1270 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 1271 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1272 // CHECK3: omp.inner.for.cond: 1273 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1274 // CHECK3-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP6]], 123 1275 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1276 // CHECK3: omp.inner.for.body: 1277 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1278 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1279 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l36.omp_outlined.omp_outlined, i32 [[TMP7]], i32 [[TMP8]], ptr [[TMP0]]) 1280 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1281 // CHECK3: omp.inner.for.inc: 1282 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1283 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 1284 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]] 1285 // CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 1286 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1287 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 1288 // CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], [[TMP12]] 1289 // CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4 1290 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1291 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 1292 // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP13]], [[TMP14]] 1293 // CHECK3-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4 1294 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1295 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP15]], 122 1296 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]] 1297 // CHECK3: cond.true5: 1298 // CHECK3-NEXT: br label [[COND_END7:%.*]] 1299 // CHECK3: cond.false6: 1300 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1301 // CHECK3-NEXT: br label [[COND_END7]] 1302 // CHECK3: cond.end7: 1303 // CHECK3-NEXT: [[COND8:%.*]] = phi i32 [ 122, [[COND_TRUE5]] ], [ [[TMP16]], [[COND_FALSE6]] ] 1304 // CHECK3-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4 1305 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1306 // CHECK3-NEXT: store i32 [[TMP17]], ptr [[DOTOMP_IV]], align 4 1307 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 1308 // CHECK3: omp.inner.for.end: 1309 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1310 // CHECK3: omp.loop.exit: 1311 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 1312 // CHECK3-NEXT: ret void 1313 // 1314 // 1315 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l36.omp_outlined.omp_outlined 1316 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 1317 // CHECK3-NEXT: entry: 1318 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 1319 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 1320 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 1321 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 1322 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 1323 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1324 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 1325 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1326 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1327 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1328 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1329 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 1330 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 1331 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 1332 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4 1333 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4 1334 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 1335 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 1336 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1337 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 1338 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4 1339 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4 1340 // CHECK3-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4 1341 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4 1342 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1343 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1344 // CHECK3-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1345 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 1346 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1347 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1348 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122 1349 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1350 // CHECK3: cond.true: 1351 // CHECK3-NEXT: br label [[COND_END:%.*]] 1352 // CHECK3: cond.false: 1353 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1354 // CHECK3-NEXT: br label [[COND_END]] 1355 // CHECK3: cond.end: 1356 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 1357 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 1358 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1359 // CHECK3-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 1360 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1361 // CHECK3: omp.inner.for.cond: 1362 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1363 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1364 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 1365 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1366 // CHECK3: omp.inner.for.body: 1367 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1368 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 1369 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1370 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4 1371 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 1372 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4 1373 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP11]] 1374 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 1375 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1376 // CHECK3: omp.body.continue: 1377 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1378 // CHECK3: omp.inner.for.inc: 1379 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1380 // CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1 1381 // CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 1382 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 1383 // CHECK3: omp.inner.for.end: 1384 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1385 // CHECK3: omp.loop.exit: 1386 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]]) 1387 // CHECK3-NEXT: ret void 1388 // 1389 // 1390 // CHECK9-LABEL: define {{[^@]+}}@main 1391 // CHECK9-SAME: (i32 noundef signext [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 1392 // CHECK9-NEXT: entry: 1393 // CHECK9-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 1394 // CHECK9-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 1395 // CHECK9-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 8 1396 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 1397 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 8 1398 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 1399 // CHECK9-NEXT: [[M:%.*]] = alloca i32, align 4 1400 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 1401 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8 1402 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8 1403 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8 1404 // CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8 1405 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1406 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1407 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1408 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 1409 // CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 1410 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x ptr], align 8 1411 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x ptr], align 8 1412 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x ptr], align 8 1413 // CHECK9-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 8 1414 // CHECK9-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 1415 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 1416 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 1417 // CHECK9-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 1418 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4 1419 // CHECK9-NEXT: [[N_CASTED19:%.*]] = alloca i64, align 8 1420 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 1421 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [4 x ptr], align 8 1422 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [4 x ptr], align 8 1423 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [4 x ptr], align 8 1424 // CHECK9-NEXT: [[DOTOFFLOAD_SIZES23:%.*]] = alloca [4 x i64], align 8 1425 // CHECK9-NEXT: [[_TMP24:%.*]] = alloca i32, align 4 1426 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4 1427 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4 1428 // CHECK9-NEXT: [[KERNEL_ARGS31:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 1429 // CHECK9-NEXT: store i32 0, ptr [[RETVAL]], align 4 1430 // CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 1431 // CHECK9-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8 1432 // CHECK9-NEXT: store i32 100, ptr [[N]], align 4 1433 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 1434 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 1435 // CHECK9-NEXT: [[TMP2:%.*]] = call ptr @llvm.stacksave.p0() 1436 // CHECK9-NEXT: store ptr [[TMP2]], ptr [[SAVED_STACK]], align 8 1437 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 1438 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8 1439 // CHECK9-NEXT: store i32 10, ptr [[M]], align 4 1440 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N]], align 4 1441 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 1442 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8 1443 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 1444 // CHECK9-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 24, i1 false) 1445 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1446 // CHECK9-NEXT: store i64 [[TMP4]], ptr [[TMP6]], align 8 1447 // CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1448 // CHECK9-NEXT: store i64 [[TMP4]], ptr [[TMP7]], align 8 1449 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 1450 // CHECK9-NEXT: store ptr null, ptr [[TMP8]], align 8 1451 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 1452 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP9]], align 8 1453 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 1454 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP10]], align 8 1455 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 1456 // CHECK9-NEXT: store ptr null, ptr [[TMP11]], align 8 1457 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 1458 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP12]], align 8 1459 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 1460 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP13]], align 8 1461 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2 1462 // CHECK9-NEXT: store i64 [[TMP5]], ptr [[TMP14]], align 8 1463 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 1464 // CHECK9-NEXT: store ptr null, ptr [[TMP15]], align 8 1465 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1466 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1467 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 1468 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, ptr [[N]], align 4 1469 // CHECK9-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4 1470 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1471 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0 1472 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1473 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1474 // CHECK9-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1475 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1476 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], 1 1477 // CHECK9-NEXT: [[TMP22:%.*]] = zext i32 [[ADD]] to i64 1478 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 1479 // CHECK9-NEXT: store i32 3, ptr [[TMP23]], align 4 1480 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 1481 // CHECK9-NEXT: store i32 3, ptr [[TMP24]], align 4 1482 // CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 1483 // CHECK9-NEXT: store ptr [[TMP16]], ptr [[TMP25]], align 8 1484 // CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 1485 // CHECK9-NEXT: store ptr [[TMP17]], ptr [[TMP26]], align 8 1486 // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 1487 // CHECK9-NEXT: store ptr [[TMP18]], ptr [[TMP27]], align 8 1488 // CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 1489 // CHECK9-NEXT: store ptr @.offload_maptypes, ptr [[TMP28]], align 8 1490 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 1491 // CHECK9-NEXT: store ptr null, ptr [[TMP29]], align 8 1492 // CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 1493 // CHECK9-NEXT: store ptr null, ptr [[TMP30]], align 8 1494 // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 1495 // CHECK9-NEXT: store i64 [[TMP22]], ptr [[TMP31]], align 8 1496 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 1497 // CHECK9-NEXT: store i64 0, ptr [[TMP32]], align 8 1498 // CHECK9-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 1499 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4 1500 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 1501 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP34]], align 4 1502 // CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 1503 // CHECK9-NEXT: store i32 0, ptr [[TMP35]], align 4 1504 // CHECK9-NEXT: [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, ptr [[KERNEL_ARGS]]) 1505 // CHECK9-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0 1506 // CHECK9-NEXT: br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1507 // CHECK9: omp_offload.failed: 1508 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3:[0-9]+]] 1509 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] 1510 // CHECK9: omp_offload.cont: 1511 // CHECK9-NEXT: [[TMP38:%.*]] = load i32, ptr [[N]], align 4 1512 // CHECK9-NEXT: store i32 [[TMP38]], ptr [[N_CASTED3]], align 4 1513 // CHECK9-NEXT: [[TMP39:%.*]] = load i64, ptr [[N_CASTED3]], align 8 1514 // CHECK9-NEXT: [[TMP40:%.*]] = mul nuw i64 [[TMP1]], 4 1515 // CHECK9-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES7]], ptr align 8 @.offload_sizes.1, i64 24, i1 false) 1516 // CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 1517 // CHECK9-NEXT: store i64 [[TMP39]], ptr [[TMP41]], align 8 1518 // CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 1519 // CHECK9-NEXT: store i64 [[TMP39]], ptr [[TMP42]], align 8 1520 // CHECK9-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i64 0, i64 0 1521 // CHECK9-NEXT: store ptr null, ptr [[TMP43]], align 8 1522 // CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 1523 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP44]], align 8 1524 // CHECK9-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 1525 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP45]], align 8 1526 // CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i64 0, i64 1 1527 // CHECK9-NEXT: store ptr null, ptr [[TMP46]], align 8 1528 // CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 1529 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP47]], align 8 1530 // CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 1531 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP48]], align 8 1532 // CHECK9-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 1533 // CHECK9-NEXT: store i64 [[TMP40]], ptr [[TMP49]], align 8 1534 // CHECK9-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i64 0, i64 2 1535 // CHECK9-NEXT: store ptr null, ptr [[TMP50]], align 8 1536 // CHECK9-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 1537 // CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 1538 // CHECK9-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 1539 // CHECK9-NEXT: [[TMP54:%.*]] = load i32, ptr [[N]], align 4 1540 // CHECK9-NEXT: store i32 [[TMP54]], ptr [[DOTCAPTURE_EXPR_9]], align 4 1541 // CHECK9-NEXT: [[TMP55:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_9]], align 4 1542 // CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP55]], 0 1543 // CHECK9-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 1544 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 1545 // CHECK9-NEXT: store i32 [[SUB13]], ptr [[DOTCAPTURE_EXPR_10]], align 4 1546 // CHECK9-NEXT: [[TMP56:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_10]], align 4 1547 // CHECK9-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP56]], 1 1548 // CHECK9-NEXT: [[TMP57:%.*]] = zext i32 [[ADD14]] to i64 1549 // CHECK9-NEXT: [[TMP58:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0 1550 // CHECK9-NEXT: store i32 3, ptr [[TMP58]], align 4 1551 // CHECK9-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1 1552 // CHECK9-NEXT: store i32 3, ptr [[TMP59]], align 4 1553 // CHECK9-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2 1554 // CHECK9-NEXT: store ptr [[TMP51]], ptr [[TMP60]], align 8 1555 // CHECK9-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3 1556 // CHECK9-NEXT: store ptr [[TMP52]], ptr [[TMP61]], align 8 1557 // CHECK9-NEXT: [[TMP62:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4 1558 // CHECK9-NEXT: store ptr [[TMP53]], ptr [[TMP62]], align 8 1559 // CHECK9-NEXT: [[TMP63:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5 1560 // CHECK9-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP63]], align 8 1561 // CHECK9-NEXT: [[TMP64:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6 1562 // CHECK9-NEXT: store ptr null, ptr [[TMP64]], align 8 1563 // CHECK9-NEXT: [[TMP65:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7 1564 // CHECK9-NEXT: store ptr null, ptr [[TMP65]], align 8 1565 // CHECK9-NEXT: [[TMP66:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8 1566 // CHECK9-NEXT: store i64 [[TMP57]], ptr [[TMP66]], align 8 1567 // CHECK9-NEXT: [[TMP67:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9 1568 // CHECK9-NEXT: store i64 0, ptr [[TMP67]], align 8 1569 // CHECK9-NEXT: [[TMP68:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10 1570 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP68]], align 4 1571 // CHECK9-NEXT: [[TMP69:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11 1572 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP69]], align 4 1573 // CHECK9-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12 1574 // CHECK9-NEXT: store i32 0, ptr [[TMP70]], align 4 1575 // CHECK9-NEXT: [[TMP71:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, ptr [[KERNEL_ARGS15]]) 1576 // CHECK9-NEXT: [[TMP72:%.*]] = icmp ne i32 [[TMP71]], 0 1577 // CHECK9-NEXT: br i1 [[TMP72]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] 1578 // CHECK9: omp_offload.failed16: 1579 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i64 [[TMP39]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3]] 1580 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT17]] 1581 // CHECK9: omp_offload.cont17: 1582 // CHECK9-NEXT: [[TMP73:%.*]] = load i32, ptr [[M]], align 4 1583 // CHECK9-NEXT: store i32 [[TMP73]], ptr [[DOTCAPTURE_EXPR_18]], align 4 1584 // CHECK9-NEXT: [[TMP74:%.*]] = load i32, ptr [[N]], align 4 1585 // CHECK9-NEXT: store i32 [[TMP74]], ptr [[N_CASTED19]], align 4 1586 // CHECK9-NEXT: [[TMP75:%.*]] = load i64, ptr [[N_CASTED19]], align 8 1587 // CHECK9-NEXT: [[TMP76:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_18]], align 4 1588 // CHECK9-NEXT: store i32 [[TMP76]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 1589 // CHECK9-NEXT: [[TMP77:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8 1590 // CHECK9-NEXT: [[TMP78:%.*]] = mul nuw i64 [[TMP1]], 4 1591 // CHECK9-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES23]], ptr align 8 @.offload_sizes.3, i64 32, i1 false) 1592 // CHECK9-NEXT: [[TMP79:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 1593 // CHECK9-NEXT: store i64 [[TMP75]], ptr [[TMP79]], align 8 1594 // CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 1595 // CHECK9-NEXT: store i64 [[TMP75]], ptr [[TMP80]], align 8 1596 // CHECK9-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 0 1597 // CHECK9-NEXT: store ptr null, ptr [[TMP81]], align 8 1598 // CHECK9-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1 1599 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP82]], align 8 1600 // CHECK9-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS21]], i32 0, i32 1 1601 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP83]], align 8 1602 // CHECK9-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 1 1603 // CHECK9-NEXT: store ptr null, ptr [[TMP84]], align 8 1604 // CHECK9-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2 1605 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP85]], align 8 1606 // CHECK9-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS21]], i32 0, i32 2 1607 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP86]], align 8 1608 // CHECK9-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES23]], i32 0, i32 2 1609 // CHECK9-NEXT: store i64 [[TMP78]], ptr [[TMP87]], align 8 1610 // CHECK9-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 2 1611 // CHECK9-NEXT: store ptr null, ptr [[TMP88]], align 8 1612 // CHECK9-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3 1613 // CHECK9-NEXT: store i64 [[TMP77]], ptr [[TMP89]], align 8 1614 // CHECK9-NEXT: [[TMP90:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS21]], i32 0, i32 3 1615 // CHECK9-NEXT: store i64 [[TMP77]], ptr [[TMP90]], align 8 1616 // CHECK9-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS22]], i64 0, i64 3 1617 // CHECK9-NEXT: store ptr null, ptr [[TMP91]], align 8 1618 // CHECK9-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 1619 // CHECK9-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 1620 // CHECK9-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES23]], i32 0, i32 0 1621 // CHECK9-NEXT: [[TMP95:%.*]] = load i32, ptr [[N]], align 4 1622 // CHECK9-NEXT: store i32 [[TMP95]], ptr [[DOTCAPTURE_EXPR_25]], align 4 1623 // CHECK9-NEXT: [[TMP96:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_25]], align 4 1624 // CHECK9-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP96]], 0 1625 // CHECK9-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1 1626 // CHECK9-NEXT: [[SUB29:%.*]] = sub nsw i32 [[DIV28]], 1 1627 // CHECK9-NEXT: store i32 [[SUB29]], ptr [[DOTCAPTURE_EXPR_26]], align 4 1628 // CHECK9-NEXT: [[TMP97:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_26]], align 4 1629 // CHECK9-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP97]], 1 1630 // CHECK9-NEXT: [[TMP98:%.*]] = zext i32 [[ADD30]] to i64 1631 // CHECK9-NEXT: [[TMP99:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 0 1632 // CHECK9-NEXT: store i32 3, ptr [[TMP99]], align 4 1633 // CHECK9-NEXT: [[TMP100:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 1 1634 // CHECK9-NEXT: store i32 4, ptr [[TMP100]], align 4 1635 // CHECK9-NEXT: [[TMP101:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 2 1636 // CHECK9-NEXT: store ptr [[TMP92]], ptr [[TMP101]], align 8 1637 // CHECK9-NEXT: [[TMP102:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 3 1638 // CHECK9-NEXT: store ptr [[TMP93]], ptr [[TMP102]], align 8 1639 // CHECK9-NEXT: [[TMP103:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 4 1640 // CHECK9-NEXT: store ptr [[TMP94]], ptr [[TMP103]], align 8 1641 // CHECK9-NEXT: [[TMP104:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 5 1642 // CHECK9-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP104]], align 8 1643 // CHECK9-NEXT: [[TMP105:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 6 1644 // CHECK9-NEXT: store ptr null, ptr [[TMP105]], align 8 1645 // CHECK9-NEXT: [[TMP106:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 7 1646 // CHECK9-NEXT: store ptr null, ptr [[TMP106]], align 8 1647 // CHECK9-NEXT: [[TMP107:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 8 1648 // CHECK9-NEXT: store i64 [[TMP98]], ptr [[TMP107]], align 8 1649 // CHECK9-NEXT: [[TMP108:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 9 1650 // CHECK9-NEXT: store i64 0, ptr [[TMP108]], align 8 1651 // CHECK9-NEXT: [[TMP109:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 10 1652 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP109]], align 4 1653 // CHECK9-NEXT: [[TMP110:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 11 1654 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP110]], align 4 1655 // CHECK9-NEXT: [[TMP111:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 12 1656 // CHECK9-NEXT: store i32 0, ptr [[TMP111]], align 4 1657 // CHECK9-NEXT: [[TMP112:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, ptr [[KERNEL_ARGS31]]) 1658 // CHECK9-NEXT: [[TMP113:%.*]] = icmp ne i32 [[TMP112]], 0 1659 // CHECK9-NEXT: br i1 [[TMP113]], label [[OMP_OFFLOAD_FAILED32:%.*]], label [[OMP_OFFLOAD_CONT33:%.*]] 1660 // CHECK9: omp_offload.failed32: 1661 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i64 [[TMP75]], i64 [[TMP1]], ptr [[VLA]], i64 [[TMP77]]) #[[ATTR3]] 1662 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT33]] 1663 // CHECK9: omp_offload.cont33: 1664 // CHECK9-NEXT: [[TMP114:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 1665 // CHECK9-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10EEiT_(i32 noundef signext [[TMP114]]) 1666 // CHECK9-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 1667 // CHECK9-NEXT: [[TMP115:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 1668 // CHECK9-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP115]]) 1669 // CHECK9-NEXT: [[TMP116:%.*]] = load i32, ptr [[RETVAL]], align 4 1670 // CHECK9-NEXT: ret i32 [[TMP116]] 1671 // 1672 // 1673 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 1674 // CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 1675 // CHECK9-NEXT: entry: 1676 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 1677 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1678 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1679 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 1680 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 1681 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 1682 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1683 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 1684 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1685 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 1686 // CHECK9-NEXT: store i32 [[TMP2]], ptr [[N_CASTED]], align 4 1687 // CHECK9-NEXT: [[TMP3:%.*]] = load i64, ptr [[N_CASTED]], align 8 1688 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.omp_outlined, i64 [[TMP3]], i64 [[TMP0]], ptr [[TMP1]]) 1689 // CHECK9-NEXT: ret void 1690 // 1691 // 1692 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.omp_outlined 1693 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1694 // CHECK9-NEXT: entry: 1695 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 1696 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 1697 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 1698 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1699 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1700 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1701 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1702 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1703 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1704 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 1705 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 1706 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 1707 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1708 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1709 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 1710 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 1711 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 1712 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 1713 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 1714 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 1715 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1716 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 1717 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1718 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 1719 // CHECK9-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 1720 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1721 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 1722 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1723 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1724 // CHECK9-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1725 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4 1726 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1727 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 1728 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1729 // CHECK9: omp.precond.then: 1730 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 1731 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1732 // CHECK9-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_COMB_UB]], align 4 1733 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1734 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1735 // CHECK9-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1736 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4 1737 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP7]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1738 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1739 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1740 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 1741 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1742 // CHECK9: cond.true: 1743 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1744 // CHECK9-NEXT: br label [[COND_END:%.*]] 1745 // CHECK9: cond.false: 1746 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1747 // CHECK9-NEXT: br label [[COND_END]] 1748 // CHECK9: cond.end: 1749 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 1750 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 1751 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1752 // CHECK9-NEXT: store i32 [[TMP12]], ptr [[DOTOMP_IV]], align 4 1753 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1754 // CHECK9: omp.inner.for.cond: 1755 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1756 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1757 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 1758 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1759 // CHECK9: omp.inner.for.body: 1760 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1761 // CHECK9-NEXT: [[TMP16:%.*]] = zext i32 [[TMP15]] to i64 1762 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1763 // CHECK9-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 1764 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, ptr [[N_ADDR]], align 4 1765 // CHECK9-NEXT: store i32 [[TMP19]], ptr [[N_CASTED]], align 4 1766 // CHECK9-NEXT: [[TMP20:%.*]] = load i64, ptr [[N_CASTED]], align 8 1767 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.omp_outlined.omp_outlined, i64 [[TMP16]], i64 [[TMP18]], i64 [[TMP20]], i64 [[TMP0]], ptr [[TMP1]]) 1768 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1769 // CHECK9: omp.inner.for.inc: 1770 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1771 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 1772 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 1773 // CHECK9-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 1774 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 1775 // CHECK9: omp.inner.for.end: 1776 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1777 // CHECK9: omp.loop.exit: 1778 // CHECK9-NEXT: [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1779 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4 1780 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP24]]) 1781 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 1782 // CHECK9: omp.precond.end: 1783 // CHECK9-NEXT: ret void 1784 // 1785 // 1786 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.omp_outlined.omp_outlined 1787 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1788 // CHECK9-NEXT: entry: 1789 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 1790 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 1791 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 1792 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 1793 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 1794 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1795 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1796 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1797 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1798 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1799 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1800 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 1801 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1802 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1803 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1804 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1805 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 1806 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 1807 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 1808 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8 1809 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8 1810 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 1811 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 1812 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1813 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 1814 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1815 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 1816 // CHECK9-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 1817 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1818 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 1819 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1820 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1821 // CHECK9-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1822 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4 1823 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1824 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 1825 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1826 // CHECK9: omp.precond.then: 1827 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1828 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1829 // CHECK9-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_UB]], align 4 1830 // CHECK9-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8 1831 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP6]] to i32 1832 // CHECK9-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 1833 // CHECK9-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP7]] to i32 1834 // CHECK9-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4 1835 // CHECK9-NEXT: store i32 [[CONV3]], ptr [[DOTOMP_UB]], align 4 1836 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1837 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1838 // CHECK9-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1839 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4 1840 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP9]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1841 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1842 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1843 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 1844 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1845 // CHECK9: cond.true: 1846 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1847 // CHECK9-NEXT: br label [[COND_END:%.*]] 1848 // CHECK9: cond.false: 1849 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1850 // CHECK9-NEXT: br label [[COND_END]] 1851 // CHECK9: cond.end: 1852 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 1853 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 1854 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1855 // CHECK9-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4 1856 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1857 // CHECK9: omp.inner.for.cond: 1858 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1859 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1860 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 1861 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1862 // CHECK9: omp.inner.for.body: 1863 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1864 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 1865 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1866 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I4]], align 4 1867 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, ptr [[I4]], align 4 1868 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP18]] to i64 1869 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 [[IDXPROM]] 1870 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 1871 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1872 // CHECK9: omp.body.continue: 1873 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1874 // CHECK9: omp.inner.for.inc: 1875 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1876 // CHECK9-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP19]], 1 1877 // CHECK9-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4 1878 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 1879 // CHECK9: omp.inner.for.end: 1880 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1881 // CHECK9: omp.loop.exit: 1882 // CHECK9-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1883 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4 1884 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP21]]) 1885 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 1886 // CHECK9: omp.precond.end: 1887 // CHECK9-NEXT: ret void 1888 // 1889 // 1890 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104 1891 // CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1892 // CHECK9-NEXT: entry: 1893 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 1894 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1895 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1896 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 1897 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 1898 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 1899 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1900 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 1901 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1902 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 1903 // CHECK9-NEXT: store i32 [[TMP2]], ptr [[N_CASTED]], align 4 1904 // CHECK9-NEXT: [[TMP3:%.*]] = load i64, ptr [[N_CASTED]], align 8 1905 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.omp_outlined, i64 [[TMP3]], i64 [[TMP0]], ptr [[TMP1]]) 1906 // CHECK9-NEXT: ret void 1907 // 1908 // 1909 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.omp_outlined 1910 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1911 // CHECK9-NEXT: entry: 1912 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 1913 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 1914 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 1915 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1916 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1917 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1918 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1919 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1920 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1921 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 1922 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 1923 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 1924 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1925 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1926 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 1927 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 1928 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 1929 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 1930 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 1931 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 1932 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1933 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 1934 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1935 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 1936 // CHECK9-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 1937 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1938 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 1939 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1940 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1941 // CHECK9-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1942 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4 1943 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1944 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 1945 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1946 // CHECK9: omp.precond.then: 1947 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 1948 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1949 // CHECK9-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_COMB_UB]], align 4 1950 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1951 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1952 // CHECK9-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1953 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4 1954 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP7]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1955 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1956 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1957 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 1958 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1959 // CHECK9: cond.true: 1960 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1961 // CHECK9-NEXT: br label [[COND_END:%.*]] 1962 // CHECK9: cond.false: 1963 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1964 // CHECK9-NEXT: br label [[COND_END]] 1965 // CHECK9: cond.end: 1966 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 1967 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 1968 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1969 // CHECK9-NEXT: store i32 [[TMP12]], ptr [[DOTOMP_IV]], align 4 1970 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1971 // CHECK9: omp.inner.for.cond: 1972 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1973 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1974 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 1975 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1976 // CHECK9: omp.inner.for.body: 1977 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1978 // CHECK9-NEXT: [[TMP16:%.*]] = zext i32 [[TMP15]] to i64 1979 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1980 // CHECK9-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 1981 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, ptr [[N_ADDR]], align 4 1982 // CHECK9-NEXT: store i32 [[TMP19]], ptr [[N_CASTED]], align 4 1983 // CHECK9-NEXT: [[TMP20:%.*]] = load i64, ptr [[N_CASTED]], align 8 1984 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.omp_outlined.omp_outlined, i64 [[TMP16]], i64 [[TMP18]], i64 [[TMP20]], i64 [[TMP0]], ptr [[TMP1]]) 1985 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1986 // CHECK9: omp.inner.for.inc: 1987 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1988 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 1989 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], [[TMP22]] 1990 // CHECK9-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 1991 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 1992 // CHECK9: omp.inner.for.end: 1993 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1994 // CHECK9: omp.loop.exit: 1995 // CHECK9-NEXT: [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1996 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4 1997 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP24]]) 1998 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 1999 // CHECK9: omp.precond.end: 2000 // CHECK9-NEXT: ret void 2001 // 2002 // 2003 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.omp_outlined.omp_outlined 2004 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 2005 // CHECK9-NEXT: entry: 2006 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 2007 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 2008 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 2009 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 2010 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 2011 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 2012 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2013 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2014 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 2015 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2016 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2017 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 2018 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2019 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2020 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2021 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2022 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 2023 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 2024 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 2025 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8 2026 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8 2027 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 2028 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 2029 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2030 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 2031 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2032 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 2033 // CHECK9-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 2034 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 2035 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 2036 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2037 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 2038 // CHECK9-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 2039 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4 2040 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 2041 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 2042 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2043 // CHECK9: omp.precond.then: 2044 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2045 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2046 // CHECK9-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_UB]], align 4 2047 // CHECK9-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8 2048 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP6]] to i32 2049 // CHECK9-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 2050 // CHECK9-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP7]] to i32 2051 // CHECK9-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4 2052 // CHECK9-NEXT: store i32 [[CONV3]], ptr [[DOTOMP_UB]], align 4 2053 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2054 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2055 // CHECK9-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2056 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4 2057 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP9]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2058 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2059 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2060 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 2061 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2062 // CHECK9: cond.true: 2063 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2064 // CHECK9-NEXT: br label [[COND_END:%.*]] 2065 // CHECK9: cond.false: 2066 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2067 // CHECK9-NEXT: br label [[COND_END]] 2068 // CHECK9: cond.end: 2069 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 2070 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2071 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2072 // CHECK9-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4 2073 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2074 // CHECK9: omp.inner.for.cond: 2075 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2076 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2077 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 2078 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2079 // CHECK9: omp.inner.for.body: 2080 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2081 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 2082 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2083 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I4]], align 4 2084 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, ptr [[I4]], align 4 2085 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP18]] to i64 2086 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 [[IDXPROM]] 2087 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 2088 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2089 // CHECK9: omp.body.continue: 2090 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2091 // CHECK9: omp.inner.for.inc: 2092 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2093 // CHECK9-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP19]], 1 2094 // CHECK9-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4 2095 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 2096 // CHECK9: omp.inner.for.end: 2097 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2098 // CHECK9: omp.loop.exit: 2099 // CHECK9-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2100 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4 2101 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP21]]) 2102 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 2103 // CHECK9: omp.precond.end: 2104 // CHECK9-NEXT: ret void 2105 // 2106 // 2107 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108 2108 // CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 2109 // CHECK9-NEXT: entry: 2110 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 2111 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 2112 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2113 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 2114 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 2115 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 2116 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 2117 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 2118 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2119 // CHECK9-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8 2120 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 2121 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2122 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 2123 // CHECK9-NEXT: store i32 [[TMP2]], ptr [[N_CASTED]], align 4 2124 // CHECK9-NEXT: [[TMP3:%.*]] = load i64, ptr [[N_CASTED]], align 8 2125 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 2126 // CHECK9-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 2127 // CHECK9-NEXT: [[TMP5:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8 2128 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.omp_outlined, i64 [[TMP3]], i64 [[TMP0]], ptr [[TMP1]], i64 [[TMP5]]) 2129 // CHECK9-NEXT: ret void 2130 // 2131 // 2132 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.omp_outlined 2133 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 2134 // CHECK9-NEXT: entry: 2135 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 2136 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 2137 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 2138 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 2139 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2140 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 2141 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2142 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 2143 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2144 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 2145 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 2146 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 2147 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 2148 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2149 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2150 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 2151 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 2152 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 2153 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 2154 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 2155 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 2156 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 2157 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2158 // CHECK9-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8 2159 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 2160 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2161 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 2162 // CHECK9-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 2163 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2164 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 2165 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2166 // CHECK9-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 2167 // CHECK9-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4 2168 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4 2169 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2170 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 2171 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2172 // CHECK9: omp.precond.then: 2173 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 2174 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 2175 // CHECK9-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_COMB_UB]], align 4 2176 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2177 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2178 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 2179 // CHECK9-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2180 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 2181 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP6]]) 2182 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2183 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 2184 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 2185 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2186 // CHECK9: cond.true: 2187 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 2188 // CHECK9-NEXT: br label [[COND_END:%.*]] 2189 // CHECK9: cond.false: 2190 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2191 // CHECK9-NEXT: br label [[COND_END]] 2192 // CHECK9: cond.end: 2193 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 2194 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 2195 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2196 // CHECK9-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 2197 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2198 // CHECK9: omp.inner.for.cond: 2199 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2200 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 2201 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP15]], 1 2202 // CHECK9-NEXT: [[CMP6:%.*]] = icmp slt i32 [[TMP14]], [[ADD]] 2203 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2204 // CHECK9: omp.inner.for.body: 2205 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2206 // CHECK9-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64 2207 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2208 // CHECK9-NEXT: [[TMP19:%.*]] = zext i32 [[TMP18]] to i64 2209 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[N_ADDR]], align 4 2210 // CHECK9-NEXT: store i32 [[TMP20]], ptr [[N_CASTED]], align 4 2211 // CHECK9-NEXT: [[TMP21:%.*]] = load i64, ptr [[N_CASTED]], align 8 2212 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 2213 // CHECK9-NEXT: store i32 [[TMP22]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 2214 // CHECK9-NEXT: [[TMP23:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8 2215 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 6, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.omp_outlined.omp_outlined, i64 [[TMP17]], i64 [[TMP19]], i64 [[TMP21]], i64 [[TMP0]], ptr [[TMP1]], i64 [[TMP23]]) 2216 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2217 // CHECK9: omp.inner.for.inc: 2218 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2219 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 2220 // CHECK9-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP24]], [[TMP25]] 2221 // CHECK9-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4 2222 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2223 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 2224 // CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP26]], [[TMP27]] 2225 // CHECK9-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_LB]], align 4 2226 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2227 // CHECK9-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 2228 // CHECK9-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP28]], [[TMP29]] 2229 // CHECK9-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_COMB_UB]], align 4 2230 // CHECK9-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2231 // CHECK9-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 2232 // CHECK9-NEXT: [[CMP10:%.*]] = icmp sgt i32 [[TMP30]], [[TMP31]] 2233 // CHECK9-NEXT: br i1 [[CMP10]], label [[COND_TRUE11:%.*]], label [[COND_FALSE12:%.*]] 2234 // CHECK9: cond.true11: 2235 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 2236 // CHECK9-NEXT: br label [[COND_END13:%.*]] 2237 // CHECK9: cond.false12: 2238 // CHECK9-NEXT: [[TMP33:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2239 // CHECK9-NEXT: br label [[COND_END13]] 2240 // CHECK9: cond.end13: 2241 // CHECK9-NEXT: [[COND14:%.*]] = phi i32 [ [[TMP32]], [[COND_TRUE11]] ], [ [[TMP33]], [[COND_FALSE12]] ] 2242 // CHECK9-NEXT: store i32 [[COND14]], ptr [[DOTOMP_COMB_UB]], align 4 2243 // CHECK9-NEXT: [[TMP34:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2244 // CHECK9-NEXT: store i32 [[TMP34]], ptr [[DOTOMP_IV]], align 4 2245 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 2246 // CHECK9: omp.inner.for.end: 2247 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2248 // CHECK9: omp.loop.exit: 2249 // CHECK9-NEXT: [[TMP35:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2250 // CHECK9-NEXT: [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4 2251 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP36]]) 2252 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 2253 // CHECK9: omp.precond.end: 2254 // CHECK9-NEXT: ret void 2255 // 2256 // 2257 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.omp_outlined.omp_outlined 2258 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 2259 // CHECK9-NEXT: entry: 2260 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 2261 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 2262 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 2263 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 2264 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 2265 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 2266 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2267 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 2268 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2269 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 2270 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2271 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 2272 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 2273 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2274 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2275 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2276 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2277 // CHECK9-NEXT: [[I5:%.*]] = alloca i32, align 4 2278 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 2279 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 2280 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8 2281 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8 2282 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 2283 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 2284 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2285 // CHECK9-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8 2286 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 2287 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2288 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 2289 // CHECK9-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 2290 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2291 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 2292 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2293 // CHECK9-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 2294 // CHECK9-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4 2295 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4 2296 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2297 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 2298 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2299 // CHECK9: omp.precond.then: 2300 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2301 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 2302 // CHECK9-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_UB]], align 4 2303 // CHECK9-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8 2304 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP6]] to i32 2305 // CHECK9-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 2306 // CHECK9-NEXT: [[CONV4:%.*]] = trunc i64 [[TMP7]] to i32 2307 // CHECK9-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4 2308 // CHECK9-NEXT: store i32 [[CONV4]], ptr [[DOTOMP_UB]], align 4 2309 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2310 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2311 // CHECK9-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2312 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4 2313 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP9]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2314 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2315 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 2316 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 2317 // CHECK9-NEXT: br i1 [[CMP6]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2318 // CHECK9: cond.true: 2319 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 2320 // CHECK9-NEXT: br label [[COND_END:%.*]] 2321 // CHECK9: cond.false: 2322 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2323 // CHECK9-NEXT: br label [[COND_END]] 2324 // CHECK9: cond.end: 2325 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 2326 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2327 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2328 // CHECK9-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4 2329 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2330 // CHECK9: omp.inner.for.cond: 2331 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2332 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2333 // CHECK9-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 2334 // CHECK9-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2335 // CHECK9: omp.inner.for.body: 2336 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2337 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 2338 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2339 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I5]], align 4 2340 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, ptr [[I5]], align 4 2341 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP18]] to i64 2342 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i64 [[IDXPROM]] 2343 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 2344 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2345 // CHECK9: omp.body.continue: 2346 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2347 // CHECK9: omp.inner.for.inc: 2348 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2349 // CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP19]], 1 2350 // CHECK9-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4 2351 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 2352 // CHECK9: omp.inner.for.end: 2353 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2354 // CHECK9: omp.loop.exit: 2355 // CHECK9-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2356 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4 2357 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP21]]) 2358 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 2359 // CHECK9: omp.precond.end: 2360 // CHECK9-NEXT: ret void 2361 // 2362 // 2363 // CHECK9-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 2364 // CHECK9-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 2365 // CHECK9-NEXT: entry: 2366 // CHECK9-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 2367 // CHECK9-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 2368 // CHECK9-NEXT: [[M:%.*]] = alloca i32, align 4 2369 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 2370 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 2371 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 2372 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 2373 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 2374 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [1 x ptr], align 8 2375 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [1 x ptr], align 8 2376 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [1 x ptr], align 8 2377 // CHECK9-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 2378 // CHECK9-NEXT: [[KERNEL_ARGS5:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 2379 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2380 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 2381 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS8:%.*]] = alloca [2 x ptr], align 8 2382 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS9:%.*]] = alloca [2 x ptr], align 8 2383 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS10:%.*]] = alloca [2 x ptr], align 8 2384 // CHECK9-NEXT: [[_TMP11:%.*]] = alloca i32, align 4 2385 // CHECK9-NEXT: [[KERNEL_ARGS12:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 2386 // CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 2387 // CHECK9-NEXT: store i32 10, ptr [[M]], align 4 2388 // CHECK9-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2389 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP0]], align 8 2390 // CHECK9-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2391 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP1]], align 8 2392 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 2393 // CHECK9-NEXT: store ptr null, ptr [[TMP2]], align 8 2394 // CHECK9-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2395 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2396 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 2397 // CHECK9-NEXT: store i32 3, ptr [[TMP5]], align 4 2398 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 2399 // CHECK9-NEXT: store i32 1, ptr [[TMP6]], align 4 2400 // CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 2401 // CHECK9-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8 2402 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 2403 // CHECK9-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8 2404 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 2405 // CHECK9-NEXT: store ptr @.offload_sizes.5, ptr [[TMP9]], align 8 2406 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 2407 // CHECK9-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP10]], align 8 2408 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 2409 // CHECK9-NEXT: store ptr null, ptr [[TMP11]], align 8 2410 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 2411 // CHECK9-NEXT: store ptr null, ptr [[TMP12]], align 8 2412 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 2413 // CHECK9-NEXT: store i64 10, ptr [[TMP13]], align 8 2414 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 2415 // CHECK9-NEXT: store i64 0, ptr [[TMP14]], align 8 2416 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 2417 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4 2418 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 2419 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4 2420 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 2421 // CHECK9-NEXT: store i32 0, ptr [[TMP17]], align 4 2422 // CHECK9-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, ptr [[KERNEL_ARGS]]) 2423 // CHECK9-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0 2424 // CHECK9-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2425 // CHECK9: omp_offload.failed: 2426 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81(ptr [[A]]) #[[ATTR3]] 2427 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] 2428 // CHECK9: omp_offload.cont: 2429 // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 2430 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP20]], align 8 2431 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 2432 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP21]], align 8 2433 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 2434 // CHECK9-NEXT: store ptr null, ptr [[TMP22]], align 8 2435 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 2436 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 2437 // CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 0 2438 // CHECK9-NEXT: store i32 3, ptr [[TMP25]], align 4 2439 // CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 1 2440 // CHECK9-NEXT: store i32 1, ptr [[TMP26]], align 4 2441 // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 2 2442 // CHECK9-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 8 2443 // CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 3 2444 // CHECK9-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 8 2445 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 4 2446 // CHECK9-NEXT: store ptr @.offload_sizes.7, ptr [[TMP29]], align 8 2447 // CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 5 2448 // CHECK9-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP30]], align 8 2449 // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 6 2450 // CHECK9-NEXT: store ptr null, ptr [[TMP31]], align 8 2451 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 7 2452 // CHECK9-NEXT: store ptr null, ptr [[TMP32]], align 8 2453 // CHECK9-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 8 2454 // CHECK9-NEXT: store i64 10, ptr [[TMP33]], align 8 2455 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 9 2456 // CHECK9-NEXT: store i64 0, ptr [[TMP34]], align 8 2457 // CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 10 2458 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4 2459 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 11 2460 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4 2461 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 12 2462 // CHECK9-NEXT: store i32 0, ptr [[TMP37]], align 4 2463 // CHECK9-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, ptr [[KERNEL_ARGS5]]) 2464 // CHECK9-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0 2465 // CHECK9-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]] 2466 // CHECK9: omp_offload.failed6: 2467 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85(ptr [[A]]) #[[ATTR3]] 2468 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT7]] 2469 // CHECK9: omp_offload.cont7: 2470 // CHECK9-NEXT: [[TMP40:%.*]] = load i32, ptr [[M]], align 4 2471 // CHECK9-NEXT: store i32 [[TMP40]], ptr [[DOTCAPTURE_EXPR_]], align 4 2472 // CHECK9-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 2473 // CHECK9-NEXT: store i32 [[TMP41]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 2474 // CHECK9-NEXT: [[TMP42:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8 2475 // CHECK9-NEXT: [[TMP43:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0 2476 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP43]], align 8 2477 // CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0 2478 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP44]], align 8 2479 // CHECK9-NEXT: [[TMP45:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i64 0, i64 0 2480 // CHECK9-NEXT: store ptr null, ptr [[TMP45]], align 8 2481 // CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 1 2482 // CHECK9-NEXT: store i64 [[TMP42]], ptr [[TMP46]], align 8 2483 // CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 1 2484 // CHECK9-NEXT: store i64 [[TMP42]], ptr [[TMP47]], align 8 2485 // CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i64 0, i64 1 2486 // CHECK9-NEXT: store ptr null, ptr [[TMP48]], align 8 2487 // CHECK9-NEXT: [[TMP49:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0 2488 // CHECK9-NEXT: [[TMP50:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0 2489 // CHECK9-NEXT: [[TMP51:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 0 2490 // CHECK9-NEXT: store i32 3, ptr [[TMP51]], align 4 2491 // CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 1 2492 // CHECK9-NEXT: store i32 2, ptr [[TMP52]], align 4 2493 // CHECK9-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 2 2494 // CHECK9-NEXT: store ptr [[TMP49]], ptr [[TMP53]], align 8 2495 // CHECK9-NEXT: [[TMP54:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 3 2496 // CHECK9-NEXT: store ptr [[TMP50]], ptr [[TMP54]], align 8 2497 // CHECK9-NEXT: [[TMP55:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 4 2498 // CHECK9-NEXT: store ptr @.offload_sizes.9, ptr [[TMP55]], align 8 2499 // CHECK9-NEXT: [[TMP56:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 5 2500 // CHECK9-NEXT: store ptr @.offload_maptypes.10, ptr [[TMP56]], align 8 2501 // CHECK9-NEXT: [[TMP57:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 6 2502 // CHECK9-NEXT: store ptr null, ptr [[TMP57]], align 8 2503 // CHECK9-NEXT: [[TMP58:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 7 2504 // CHECK9-NEXT: store ptr null, ptr [[TMP58]], align 8 2505 // CHECK9-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 8 2506 // CHECK9-NEXT: store i64 10, ptr [[TMP59]], align 8 2507 // CHECK9-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 9 2508 // CHECK9-NEXT: store i64 0, ptr [[TMP60]], align 8 2509 // CHECK9-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 10 2510 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP61]], align 4 2511 // CHECK9-NEXT: [[TMP62:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 11 2512 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP62]], align 4 2513 // CHECK9-NEXT: [[TMP63:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 12 2514 // CHECK9-NEXT: store i32 0, ptr [[TMP63]], align 4 2515 // CHECK9-NEXT: [[TMP64:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, ptr [[KERNEL_ARGS12]]) 2516 // CHECK9-NEXT: [[TMP65:%.*]] = icmp ne i32 [[TMP64]], 0 2517 // CHECK9-NEXT: br i1 [[TMP65]], label [[OMP_OFFLOAD_FAILED13:%.*]], label [[OMP_OFFLOAD_CONT14:%.*]] 2518 // CHECK9: omp_offload.failed13: 2519 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89(ptr [[A]], i64 [[TMP42]]) #[[ATTR3]] 2520 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT14]] 2521 // CHECK9: omp_offload.cont14: 2522 // CHECK9-NEXT: ret i32 0 2523 // 2524 // 2525 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81 2526 // CHECK9-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2527 // CHECK9-NEXT: entry: 2528 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2529 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2530 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2531 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.omp_outlined, ptr [[TMP0]]) 2532 // CHECK9-NEXT: ret void 2533 // 2534 // 2535 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.omp_outlined 2536 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2537 // CHECK9-NEXT: entry: 2538 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 2539 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 2540 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2541 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2542 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 2543 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 2544 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 2545 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2546 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2547 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 2548 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 2549 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 2550 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2551 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2552 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 2553 // CHECK9-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4 2554 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2555 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2556 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2557 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 2558 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2559 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2560 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 2561 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2562 // CHECK9: cond.true: 2563 // CHECK9-NEXT: br label [[COND_END:%.*]] 2564 // CHECK9: cond.false: 2565 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2566 // CHECK9-NEXT: br label [[COND_END]] 2567 // CHECK9: cond.end: 2568 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2569 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 2570 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2571 // CHECK9-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 2572 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2573 // CHECK9: omp.inner.for.cond: 2574 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2575 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2576 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 2577 // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2578 // CHECK9: omp.inner.for.body: 2579 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2580 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 2581 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2582 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 2583 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]]) 2584 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2585 // CHECK9: omp.inner.for.inc: 2586 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2587 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 2588 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] 2589 // CHECK9-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 2590 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 2591 // CHECK9: omp.inner.for.end: 2592 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2593 // CHECK9: omp.loop.exit: 2594 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 2595 // CHECK9-NEXT: ret void 2596 // 2597 // 2598 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.omp_outlined.omp_outlined 2599 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2600 // CHECK9-NEXT: entry: 2601 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 2602 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 2603 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 2604 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 2605 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2606 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2607 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 2608 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2609 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2610 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2611 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2612 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 2613 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 2614 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 2615 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8 2616 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8 2617 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2618 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2619 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2620 // CHECK9-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 2621 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8 2622 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32 2623 // CHECK9-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 2624 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32 2625 // CHECK9-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4 2626 // CHECK9-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4 2627 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2628 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2629 // CHECK9-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2630 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 2631 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2632 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2633 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9 2634 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2635 // CHECK9: cond.true: 2636 // CHECK9-NEXT: br label [[COND_END:%.*]] 2637 // CHECK9: cond.false: 2638 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2639 // CHECK9-NEXT: br label [[COND_END]] 2640 // CHECK9: cond.end: 2641 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 2642 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2643 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2644 // CHECK9-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 2645 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2646 // CHECK9: omp.inner.for.cond: 2647 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2648 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2649 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 2650 // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2651 // CHECK9: omp.inner.for.body: 2652 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2653 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 2654 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2655 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4 2656 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4 2657 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 2658 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]] 2659 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 2660 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2661 // CHECK9: omp.body.continue: 2662 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2663 // CHECK9: omp.inner.for.inc: 2664 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2665 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1 2666 // CHECK9-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4 2667 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 2668 // CHECK9: omp.inner.for.end: 2669 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2670 // CHECK9: omp.loop.exit: 2671 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]]) 2672 // CHECK9-NEXT: ret void 2673 // 2674 // 2675 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85 2676 // CHECK9-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2677 // CHECK9-NEXT: entry: 2678 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2679 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2680 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2681 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.omp_outlined, ptr [[TMP0]]) 2682 // CHECK9-NEXT: ret void 2683 // 2684 // 2685 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.omp_outlined 2686 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2687 // CHECK9-NEXT: entry: 2688 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 2689 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 2690 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2691 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2692 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 2693 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 2694 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 2695 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2696 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2697 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 2698 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 2699 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 2700 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2701 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2702 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 2703 // CHECK9-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4 2704 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2705 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2706 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2707 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 2708 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2709 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2710 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 2711 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2712 // CHECK9: cond.true: 2713 // CHECK9-NEXT: br label [[COND_END:%.*]] 2714 // CHECK9: cond.false: 2715 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2716 // CHECK9-NEXT: br label [[COND_END]] 2717 // CHECK9: cond.end: 2718 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2719 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 2720 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2721 // CHECK9-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 2722 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2723 // CHECK9: omp.inner.for.cond: 2724 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2725 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2726 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 2727 // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2728 // CHECK9: omp.inner.for.body: 2729 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2730 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 2731 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2732 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 2733 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]]) 2734 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2735 // CHECK9: omp.inner.for.inc: 2736 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2737 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 2738 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] 2739 // CHECK9-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 2740 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 2741 // CHECK9: omp.inner.for.end: 2742 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2743 // CHECK9: omp.loop.exit: 2744 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 2745 // CHECK9-NEXT: ret void 2746 // 2747 // 2748 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.omp_outlined.omp_outlined 2749 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2750 // CHECK9-NEXT: entry: 2751 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 2752 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 2753 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 2754 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 2755 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2756 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2757 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 2758 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2759 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2760 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2761 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2762 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 2763 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 2764 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 2765 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8 2766 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8 2767 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2768 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2769 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2770 // CHECK9-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 2771 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8 2772 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32 2773 // CHECK9-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 2774 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32 2775 // CHECK9-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4 2776 // CHECK9-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4 2777 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2778 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2779 // CHECK9-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2780 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 2781 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2782 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2783 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9 2784 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2785 // CHECK9: cond.true: 2786 // CHECK9-NEXT: br label [[COND_END:%.*]] 2787 // CHECK9: cond.false: 2788 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2789 // CHECK9-NEXT: br label [[COND_END]] 2790 // CHECK9: cond.end: 2791 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 2792 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2793 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2794 // CHECK9-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 2795 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2796 // CHECK9: omp.inner.for.cond: 2797 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2798 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2799 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 2800 // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2801 // CHECK9: omp.inner.for.body: 2802 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2803 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 2804 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2805 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4 2806 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4 2807 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 2808 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]] 2809 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 2810 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2811 // CHECK9: omp.body.continue: 2812 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2813 // CHECK9: omp.inner.for.inc: 2814 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2815 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1 2816 // CHECK9-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4 2817 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 2818 // CHECK9: omp.inner.for.end: 2819 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2820 // CHECK9: omp.loop.exit: 2821 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]]) 2822 // CHECK9-NEXT: ret void 2823 // 2824 // 2825 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89 2826 // CHECK9-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 2827 // CHECK9-NEXT: entry: 2828 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2829 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 2830 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 2831 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2832 // CHECK9-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8 2833 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2834 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 2835 // CHECK9-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 2836 // CHECK9-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8 2837 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined, ptr [[TMP0]], i64 [[TMP2]]) 2838 // CHECK9-NEXT: ret void 2839 // 2840 // 2841 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined 2842 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 2843 // CHECK9-NEXT: entry: 2844 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 2845 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 2846 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2847 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 2848 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2849 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 2850 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 2851 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 2852 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2853 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2854 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 2855 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 2856 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 2857 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 2858 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2859 // CHECK9-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8 2860 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2861 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 2862 // CHECK9-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4 2863 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2864 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2865 // CHECK9-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 2866 // CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2867 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4 2868 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP3]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP1]]) 2869 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2870 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9 2871 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2872 // CHECK9: cond.true: 2873 // CHECK9-NEXT: br label [[COND_END:%.*]] 2874 // CHECK9: cond.false: 2875 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2876 // CHECK9-NEXT: br label [[COND_END]] 2877 // CHECK9: cond.end: 2878 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] 2879 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 2880 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2881 // CHECK9-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4 2882 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2883 // CHECK9: omp.inner.for.cond: 2884 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2885 // CHECK9-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10 2886 // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2887 // CHECK9: omp.inner.for.body: 2888 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2889 // CHECK9-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 2890 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2891 // CHECK9-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 2892 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 2893 // CHECK9-NEXT: store i32 [[TMP12]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 2894 // CHECK9-NEXT: [[TMP13:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8 2895 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]], i64 [[TMP13]]) 2896 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2897 // CHECK9: omp.inner.for.inc: 2898 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2899 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 2900 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] 2901 // CHECK9-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 2902 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2903 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 2904 // CHECK9-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP16]], [[TMP17]] 2905 // CHECK9-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4 2906 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2907 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 2908 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] 2909 // CHECK9-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4 2910 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2911 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP20]], 9 2912 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]] 2913 // CHECK9: cond.true5: 2914 // CHECK9-NEXT: br label [[COND_END7:%.*]] 2915 // CHECK9: cond.false6: 2916 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2917 // CHECK9-NEXT: br label [[COND_END7]] 2918 // CHECK9: cond.end7: 2919 // CHECK9-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP21]], [[COND_FALSE6]] ] 2920 // CHECK9-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4 2921 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2922 // CHECK9-NEXT: store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4 2923 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 2924 // CHECK9: omp.inner.for.end: 2925 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2926 // CHECK9: omp.loop.exit: 2927 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP3]]) 2928 // CHECK9-NEXT: ret void 2929 // 2930 // 2931 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined.omp_outlined 2932 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 2933 // CHECK9-NEXT: entry: 2934 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 2935 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 2936 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 2937 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 2938 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2939 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 2940 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2941 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 2942 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2943 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2944 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2945 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2946 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 2947 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 2948 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 2949 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8 2950 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8 2951 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2952 // CHECK9-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8 2953 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2954 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2955 // CHECK9-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 2956 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8 2957 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32 2958 // CHECK9-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 2959 // CHECK9-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32 2960 // CHECK9-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4 2961 // CHECK9-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4 2962 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2963 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2964 // CHECK9-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2965 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 2966 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2967 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2968 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9 2969 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2970 // CHECK9: cond.true: 2971 // CHECK9-NEXT: br label [[COND_END:%.*]] 2972 // CHECK9: cond.false: 2973 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2974 // CHECK9-NEXT: br label [[COND_END]] 2975 // CHECK9: cond.end: 2976 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 2977 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2978 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2979 // CHECK9-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 2980 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2981 // CHECK9: omp.inner.for.cond: 2982 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2983 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2984 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 2985 // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2986 // CHECK9: omp.inner.for.body: 2987 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2988 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 2989 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2990 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4 2991 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4 2992 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 2993 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]] 2994 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 2995 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2996 // CHECK9: omp.body.continue: 2997 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2998 // CHECK9: omp.inner.for.inc: 2999 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3000 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1 3001 // CHECK9-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4 3002 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 3003 // CHECK9: omp.inner.for.end: 3004 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3005 // CHECK9: omp.loop.exit: 3006 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]]) 3007 // CHECK9-NEXT: ret void 3008 // 3009 // 3010 // CHECK11-LABEL: define {{[^@]+}}@main 3011 // CHECK11-SAME: (i32 noundef [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 3012 // CHECK11-NEXT: entry: 3013 // CHECK11-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 3014 // CHECK11-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 3015 // CHECK11-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 4 3016 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 3017 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 4 3018 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 3019 // CHECK11-NEXT: [[M:%.*]] = alloca i32, align 4 3020 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 3021 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4 3022 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4 3023 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4 3024 // CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4 3025 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 3026 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3027 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3028 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 3029 // CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 3030 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x ptr], align 4 3031 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x ptr], align 4 3032 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x ptr], align 4 3033 // CHECK11-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 3034 // CHECK11-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 3035 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 3036 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 3037 // CHECK11-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 3038 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_18:%.*]] = alloca i32, align 4 3039 // CHECK11-NEXT: [[N_CASTED19:%.*]] = alloca i32, align 4 3040 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 3041 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS20:%.*]] = alloca [4 x ptr], align 4 3042 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS21:%.*]] = alloca [4 x ptr], align 4 3043 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS22:%.*]] = alloca [4 x ptr], align 4 3044 // CHECK11-NEXT: [[DOTOFFLOAD_SIZES23:%.*]] = alloca [4 x i64], align 4 3045 // CHECK11-NEXT: [[_TMP24:%.*]] = alloca i32, align 4 3046 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4 3047 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_26:%.*]] = alloca i32, align 4 3048 // CHECK11-NEXT: [[KERNEL_ARGS31:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 3049 // CHECK11-NEXT: store i32 0, ptr [[RETVAL]], align 4 3050 // CHECK11-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 3051 // CHECK11-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 4 3052 // CHECK11-NEXT: store i32 100, ptr [[N]], align 4 3053 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 3054 // CHECK11-NEXT: [[TMP1:%.*]] = call ptr @llvm.stacksave.p0() 3055 // CHECK11-NEXT: store ptr [[TMP1]], ptr [[SAVED_STACK]], align 4 3056 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 3057 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[__VLA_EXPR0]], align 4 3058 // CHECK11-NEXT: store i32 10, ptr [[M]], align 4 3059 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[N]], align 4 3060 // CHECK11-NEXT: store i32 [[TMP2]], ptr [[N_CASTED]], align 4 3061 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_CASTED]], align 4 3062 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 3063 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 3064 // CHECK11-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 24, i1 false) 3065 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3066 // CHECK11-NEXT: store i32 [[TMP3]], ptr [[TMP6]], align 4 3067 // CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3068 // CHECK11-NEXT: store i32 [[TMP3]], ptr [[TMP7]], align 4 3069 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 3070 // CHECK11-NEXT: store ptr null, ptr [[TMP8]], align 4 3071 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 3072 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP9]], align 4 3073 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 3074 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP10]], align 4 3075 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 3076 // CHECK11-NEXT: store ptr null, ptr [[TMP11]], align 4 3077 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 3078 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP12]], align 4 3079 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 3080 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP13]], align 4 3081 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2 3082 // CHECK11-NEXT: store i64 [[TMP5]], ptr [[TMP14]], align 4 3083 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 3084 // CHECK11-NEXT: store ptr null, ptr [[TMP15]], align 4 3085 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3086 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3087 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 3088 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, ptr [[N]], align 4 3089 // CHECK11-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4 3090 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3091 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0 3092 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3093 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3094 // CHECK11-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 3095 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3096 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], 1 3097 // CHECK11-NEXT: [[TMP22:%.*]] = zext i32 [[ADD]] to i64 3098 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 3099 // CHECK11-NEXT: store i32 3, ptr [[TMP23]], align 4 3100 // CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 3101 // CHECK11-NEXT: store i32 3, ptr [[TMP24]], align 4 3102 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 3103 // CHECK11-NEXT: store ptr [[TMP16]], ptr [[TMP25]], align 4 3104 // CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 3105 // CHECK11-NEXT: store ptr [[TMP17]], ptr [[TMP26]], align 4 3106 // CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 3107 // CHECK11-NEXT: store ptr [[TMP18]], ptr [[TMP27]], align 4 3108 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 3109 // CHECK11-NEXT: store ptr @.offload_maptypes, ptr [[TMP28]], align 4 3110 // CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 3111 // CHECK11-NEXT: store ptr null, ptr [[TMP29]], align 4 3112 // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 3113 // CHECK11-NEXT: store ptr null, ptr [[TMP30]], align 4 3114 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 3115 // CHECK11-NEXT: store i64 [[TMP22]], ptr [[TMP31]], align 8 3116 // CHECK11-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 3117 // CHECK11-NEXT: store i64 0, ptr [[TMP32]], align 8 3118 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 3119 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4 3120 // CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 3121 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP34]], align 4 3122 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 3123 // CHECK11-NEXT: store i32 0, ptr [[TMP35]], align 4 3124 // CHECK11-NEXT: [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, ptr [[KERNEL_ARGS]]) 3125 // CHECK11-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0 3126 // CHECK11-NEXT: br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 3127 // CHECK11: omp_offload.failed: 3128 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3:[0-9]+]] 3129 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] 3130 // CHECK11: omp_offload.cont: 3131 // CHECK11-NEXT: [[TMP38:%.*]] = load i32, ptr [[N]], align 4 3132 // CHECK11-NEXT: store i32 [[TMP38]], ptr [[N_CASTED3]], align 4 3133 // CHECK11-NEXT: [[TMP39:%.*]] = load i32, ptr [[N_CASTED3]], align 4 3134 // CHECK11-NEXT: [[TMP40:%.*]] = mul nuw i32 [[TMP0]], 4 3135 // CHECK11-NEXT: [[TMP41:%.*]] = sext i32 [[TMP40]] to i64 3136 // CHECK11-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES7]], ptr align 4 @.offload_sizes.1, i32 24, i1 false) 3137 // CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 3138 // CHECK11-NEXT: store i32 [[TMP39]], ptr [[TMP42]], align 4 3139 // CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 3140 // CHECK11-NEXT: store i32 [[TMP39]], ptr [[TMP43]], align 4 3141 // CHECK11-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 3142 // CHECK11-NEXT: store ptr null, ptr [[TMP44]], align 4 3143 // CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 3144 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP45]], align 4 3145 // CHECK11-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 3146 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP46]], align 4 3147 // CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 3148 // CHECK11-NEXT: store ptr null, ptr [[TMP47]], align 4 3149 // CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 3150 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP48]], align 4 3151 // CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 3152 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP49]], align 4 3153 // CHECK11-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 3154 // CHECK11-NEXT: store i64 [[TMP41]], ptr [[TMP50]], align 4 3155 // CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 3156 // CHECK11-NEXT: store ptr null, ptr [[TMP51]], align 4 3157 // CHECK11-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 3158 // CHECK11-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 3159 // CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 3160 // CHECK11-NEXT: [[TMP55:%.*]] = load i32, ptr [[N]], align 4 3161 // CHECK11-NEXT: store i32 [[TMP55]], ptr [[DOTCAPTURE_EXPR_9]], align 4 3162 // CHECK11-NEXT: [[TMP56:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_9]], align 4 3163 // CHECK11-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP56]], 0 3164 // CHECK11-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 3165 // CHECK11-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 3166 // CHECK11-NEXT: store i32 [[SUB13]], ptr [[DOTCAPTURE_EXPR_10]], align 4 3167 // CHECK11-NEXT: [[TMP57:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_10]], align 4 3168 // CHECK11-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP57]], 1 3169 // CHECK11-NEXT: [[TMP58:%.*]] = zext i32 [[ADD14]] to i64 3170 // CHECK11-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0 3171 // CHECK11-NEXT: store i32 3, ptr [[TMP59]], align 4 3172 // CHECK11-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1 3173 // CHECK11-NEXT: store i32 3, ptr [[TMP60]], align 4 3174 // CHECK11-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2 3175 // CHECK11-NEXT: store ptr [[TMP52]], ptr [[TMP61]], align 4 3176 // CHECK11-NEXT: [[TMP62:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3 3177 // CHECK11-NEXT: store ptr [[TMP53]], ptr [[TMP62]], align 4 3178 // CHECK11-NEXT: [[TMP63:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4 3179 // CHECK11-NEXT: store ptr [[TMP54]], ptr [[TMP63]], align 4 3180 // CHECK11-NEXT: [[TMP64:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5 3181 // CHECK11-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP64]], align 4 3182 // CHECK11-NEXT: [[TMP65:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6 3183 // CHECK11-NEXT: store ptr null, ptr [[TMP65]], align 4 3184 // CHECK11-NEXT: [[TMP66:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7 3185 // CHECK11-NEXT: store ptr null, ptr [[TMP66]], align 4 3186 // CHECK11-NEXT: [[TMP67:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8 3187 // CHECK11-NEXT: store i64 [[TMP58]], ptr [[TMP67]], align 8 3188 // CHECK11-NEXT: [[TMP68:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9 3189 // CHECK11-NEXT: store i64 0, ptr [[TMP68]], align 8 3190 // CHECK11-NEXT: [[TMP69:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10 3191 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP69]], align 4 3192 // CHECK11-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11 3193 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP70]], align 4 3194 // CHECK11-NEXT: [[TMP71:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12 3195 // CHECK11-NEXT: store i32 0, ptr [[TMP71]], align 4 3196 // CHECK11-NEXT: [[TMP72:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.region_id, ptr [[KERNEL_ARGS15]]) 3197 // CHECK11-NEXT: [[TMP73:%.*]] = icmp ne i32 [[TMP72]], 0 3198 // CHECK11-NEXT: br i1 [[TMP73]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] 3199 // CHECK11: omp_offload.failed16: 3200 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104(i32 [[TMP39]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3]] 3201 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT17]] 3202 // CHECK11: omp_offload.cont17: 3203 // CHECK11-NEXT: [[TMP74:%.*]] = load i32, ptr [[M]], align 4 3204 // CHECK11-NEXT: store i32 [[TMP74]], ptr [[DOTCAPTURE_EXPR_18]], align 4 3205 // CHECK11-NEXT: [[TMP75:%.*]] = load i32, ptr [[N]], align 4 3206 // CHECK11-NEXT: store i32 [[TMP75]], ptr [[N_CASTED19]], align 4 3207 // CHECK11-NEXT: [[TMP76:%.*]] = load i32, ptr [[N_CASTED19]], align 4 3208 // CHECK11-NEXT: [[TMP77:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_18]], align 4 3209 // CHECK11-NEXT: store i32 [[TMP77]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 3210 // CHECK11-NEXT: [[TMP78:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 3211 // CHECK11-NEXT: [[TMP79:%.*]] = mul nuw i32 [[TMP0]], 4 3212 // CHECK11-NEXT: [[TMP80:%.*]] = sext i32 [[TMP79]] to i64 3213 // CHECK11-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES23]], ptr align 4 @.offload_sizes.3, i32 32, i1 false) 3214 // CHECK11-NEXT: [[TMP81:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 3215 // CHECK11-NEXT: store i32 [[TMP76]], ptr [[TMP81]], align 4 3216 // CHECK11-NEXT: [[TMP82:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 3217 // CHECK11-NEXT: store i32 [[TMP76]], ptr [[TMP82]], align 4 3218 // CHECK11-NEXT: [[TMP83:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS22]], i32 0, i32 0 3219 // CHECK11-NEXT: store ptr null, ptr [[TMP83]], align 4 3220 // CHECK11-NEXT: [[TMP84:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 1 3221 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP84]], align 4 3222 // CHECK11-NEXT: [[TMP85:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS21]], i32 0, i32 1 3223 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP85]], align 4 3224 // CHECK11-NEXT: [[TMP86:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS22]], i32 0, i32 1 3225 // CHECK11-NEXT: store ptr null, ptr [[TMP86]], align 4 3226 // CHECK11-NEXT: [[TMP87:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 2 3227 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP87]], align 4 3228 // CHECK11-NEXT: [[TMP88:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS21]], i32 0, i32 2 3229 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP88]], align 4 3230 // CHECK11-NEXT: [[TMP89:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES23]], i32 0, i32 2 3231 // CHECK11-NEXT: store i64 [[TMP80]], ptr [[TMP89]], align 4 3232 // CHECK11-NEXT: [[TMP90:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS22]], i32 0, i32 2 3233 // CHECK11-NEXT: store ptr null, ptr [[TMP90]], align 4 3234 // CHECK11-NEXT: [[TMP91:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 3 3235 // CHECK11-NEXT: store i32 [[TMP78]], ptr [[TMP91]], align 4 3236 // CHECK11-NEXT: [[TMP92:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS21]], i32 0, i32 3 3237 // CHECK11-NEXT: store i32 [[TMP78]], ptr [[TMP92]], align 4 3238 // CHECK11-NEXT: [[TMP93:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS22]], i32 0, i32 3 3239 // CHECK11-NEXT: store ptr null, ptr [[TMP93]], align 4 3240 // CHECK11-NEXT: [[TMP94:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS20]], i32 0, i32 0 3241 // CHECK11-NEXT: [[TMP95:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS21]], i32 0, i32 0 3242 // CHECK11-NEXT: [[TMP96:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES23]], i32 0, i32 0 3243 // CHECK11-NEXT: [[TMP97:%.*]] = load i32, ptr [[N]], align 4 3244 // CHECK11-NEXT: store i32 [[TMP97]], ptr [[DOTCAPTURE_EXPR_25]], align 4 3245 // CHECK11-NEXT: [[TMP98:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_25]], align 4 3246 // CHECK11-NEXT: [[SUB27:%.*]] = sub nsw i32 [[TMP98]], 0 3247 // CHECK11-NEXT: [[DIV28:%.*]] = sdiv i32 [[SUB27]], 1 3248 // CHECK11-NEXT: [[SUB29:%.*]] = sub nsw i32 [[DIV28]], 1 3249 // CHECK11-NEXT: store i32 [[SUB29]], ptr [[DOTCAPTURE_EXPR_26]], align 4 3250 // CHECK11-NEXT: [[TMP99:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_26]], align 4 3251 // CHECK11-NEXT: [[ADD30:%.*]] = add nsw i32 [[TMP99]], 1 3252 // CHECK11-NEXT: [[TMP100:%.*]] = zext i32 [[ADD30]] to i64 3253 // CHECK11-NEXT: [[TMP101:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 0 3254 // CHECK11-NEXT: store i32 3, ptr [[TMP101]], align 4 3255 // CHECK11-NEXT: [[TMP102:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 1 3256 // CHECK11-NEXT: store i32 4, ptr [[TMP102]], align 4 3257 // CHECK11-NEXT: [[TMP103:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 2 3258 // CHECK11-NEXT: store ptr [[TMP94]], ptr [[TMP103]], align 4 3259 // CHECK11-NEXT: [[TMP104:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 3 3260 // CHECK11-NEXT: store ptr [[TMP95]], ptr [[TMP104]], align 4 3261 // CHECK11-NEXT: [[TMP105:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 4 3262 // CHECK11-NEXT: store ptr [[TMP96]], ptr [[TMP105]], align 4 3263 // CHECK11-NEXT: [[TMP106:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 5 3264 // CHECK11-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP106]], align 4 3265 // CHECK11-NEXT: [[TMP107:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 6 3266 // CHECK11-NEXT: store ptr null, ptr [[TMP107]], align 4 3267 // CHECK11-NEXT: [[TMP108:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 7 3268 // CHECK11-NEXT: store ptr null, ptr [[TMP108]], align 4 3269 // CHECK11-NEXT: [[TMP109:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 8 3270 // CHECK11-NEXT: store i64 [[TMP100]], ptr [[TMP109]], align 8 3271 // CHECK11-NEXT: [[TMP110:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 9 3272 // CHECK11-NEXT: store i64 0, ptr [[TMP110]], align 8 3273 // CHECK11-NEXT: [[TMP111:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 10 3274 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP111]], align 4 3275 // CHECK11-NEXT: [[TMP112:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 11 3276 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP112]], align 4 3277 // CHECK11-NEXT: [[TMP113:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS31]], i32 0, i32 12 3278 // CHECK11-NEXT: store i32 0, ptr [[TMP113]], align 4 3279 // CHECK11-NEXT: [[TMP114:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.region_id, ptr [[KERNEL_ARGS31]]) 3280 // CHECK11-NEXT: [[TMP115:%.*]] = icmp ne i32 [[TMP114]], 0 3281 // CHECK11-NEXT: br i1 [[TMP115]], label [[OMP_OFFLOAD_FAILED32:%.*]], label [[OMP_OFFLOAD_CONT33:%.*]] 3282 // CHECK11: omp_offload.failed32: 3283 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108(i32 [[TMP76]], i32 [[TMP0]], ptr [[VLA]], i32 [[TMP78]]) #[[ATTR3]] 3284 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT33]] 3285 // CHECK11: omp_offload.cont33: 3286 // CHECK11-NEXT: [[TMP116:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 3287 // CHECK11-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10EEiT_(i32 noundef [[TMP116]]) 3288 // CHECK11-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 3289 // CHECK11-NEXT: [[TMP117:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4 3290 // CHECK11-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP117]]) 3291 // CHECK11-NEXT: [[TMP118:%.*]] = load i32, ptr [[RETVAL]], align 4 3292 // CHECK11-NEXT: ret i32 [[TMP118]] 3293 // 3294 // 3295 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 3296 // CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 3297 // CHECK11-NEXT: entry: 3298 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 3299 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 3300 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3301 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 3302 // CHECK11-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 3303 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 3304 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3305 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 3306 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3307 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 3308 // CHECK11-NEXT: store i32 [[TMP2]], ptr [[N_CASTED]], align 4 3309 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_CASTED]], align 4 3310 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.omp_outlined, i32 [[TMP3]], i32 [[TMP0]], ptr [[TMP1]]) 3311 // CHECK11-NEXT: ret void 3312 // 3313 // 3314 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.omp_outlined 3315 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 3316 // CHECK11-NEXT: entry: 3317 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 3318 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 3319 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 3320 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 3321 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3322 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3323 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 3324 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3325 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3326 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 3327 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 3328 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 3329 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3330 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3331 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 3332 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 3333 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 3334 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 3335 // CHECK11-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 3336 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 3337 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3338 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 3339 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3340 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 3341 // CHECK11-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 3342 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3343 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 3344 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3345 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3346 // CHECK11-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 3347 // CHECK11-NEXT: store i32 0, ptr [[I]], align 4 3348 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3349 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 3350 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3351 // CHECK11: omp.precond.then: 3352 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 3353 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3354 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_COMB_UB]], align 4 3355 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 3356 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 3357 // CHECK11-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3358 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4 3359 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP7]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 3360 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3361 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3362 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 3363 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3364 // CHECK11: cond.true: 3365 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3366 // CHECK11-NEXT: br label [[COND_END:%.*]] 3367 // CHECK11: cond.false: 3368 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3369 // CHECK11-NEXT: br label [[COND_END]] 3370 // CHECK11: cond.end: 3371 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 3372 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 3373 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 3374 // CHECK11-NEXT: store i32 [[TMP12]], ptr [[DOTOMP_IV]], align 4 3375 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3376 // CHECK11: omp.inner.for.cond: 3377 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3378 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3379 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 3380 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3381 // CHECK11: omp.inner.for.body: 3382 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 3383 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3384 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, ptr [[N_ADDR]], align 4 3385 // CHECK11-NEXT: store i32 [[TMP17]], ptr [[N_CASTED]], align 4 3386 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, ptr [[N_CASTED]], align 4 3387 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.omp_outlined.omp_outlined, i32 [[TMP15]], i32 [[TMP16]], i32 [[TMP18]], i32 [[TMP0]], ptr [[TMP1]]) 3388 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3389 // CHECK11: omp.inner.for.inc: 3390 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3391 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 3392 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 3393 // CHECK11-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 3394 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 3395 // CHECK11: omp.inner.for.end: 3396 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3397 // CHECK11: omp.loop.exit: 3398 // CHECK11-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3399 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4 3400 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP22]]) 3401 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 3402 // CHECK11: omp.precond.end: 3403 // CHECK11-NEXT: ret void 3404 // 3405 // 3406 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.omp_outlined.omp_outlined 3407 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 3408 // CHECK11-NEXT: entry: 3409 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 3410 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 3411 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 3412 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 3413 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 3414 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 3415 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3416 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3417 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 3418 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3419 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3420 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 3421 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3422 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3423 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3424 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3425 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 3426 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 3427 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 3428 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4 3429 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4 3430 // CHECK11-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 3431 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 3432 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3433 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 3434 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3435 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 3436 // CHECK11-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 3437 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3438 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 3439 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3440 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3441 // CHECK11-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 3442 // CHECK11-NEXT: store i32 0, ptr [[I]], align 4 3443 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3444 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 3445 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3446 // CHECK11: omp.precond.then: 3447 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 3448 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3449 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_UB]], align 4 3450 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4 3451 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4 3452 // CHECK11-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_LB]], align 4 3453 // CHECK11-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_UB]], align 4 3454 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 3455 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 3456 // CHECK11-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3457 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4 3458 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP9]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 3459 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3460 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3461 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 3462 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3463 // CHECK11: cond.true: 3464 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3465 // CHECK11-NEXT: br label [[COND_END:%.*]] 3466 // CHECK11: cond.false: 3467 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3468 // CHECK11-NEXT: br label [[COND_END]] 3469 // CHECK11: cond.end: 3470 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 3471 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 3472 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 3473 // CHECK11-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4 3474 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3475 // CHECK11: omp.inner.for.cond: 3476 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3477 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3478 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 3479 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3480 // CHECK11: omp.inner.for.body: 3481 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3482 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 3483 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3484 // CHECK11-NEXT: store i32 [[ADD]], ptr [[I3]], align 4 3485 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, ptr [[I3]], align 4 3486 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 [[TMP18]] 3487 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 3488 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3489 // CHECK11: omp.body.continue: 3490 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3491 // CHECK11: omp.inner.for.inc: 3492 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3493 // CHECK11-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP19]], 1 3494 // CHECK11-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 3495 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 3496 // CHECK11: omp.inner.for.end: 3497 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3498 // CHECK11: omp.loop.exit: 3499 // CHECK11-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3500 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4 3501 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP21]]) 3502 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 3503 // CHECK11: omp.precond.end: 3504 // CHECK11-NEXT: ret void 3505 // 3506 // 3507 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104 3508 // CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 3509 // CHECK11-NEXT: entry: 3510 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 3511 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 3512 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3513 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 3514 // CHECK11-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 3515 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 3516 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3517 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 3518 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3519 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 3520 // CHECK11-NEXT: store i32 [[TMP2]], ptr [[N_CASTED]], align 4 3521 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_CASTED]], align 4 3522 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.omp_outlined, i32 [[TMP3]], i32 [[TMP0]], ptr [[TMP1]]) 3523 // CHECK11-NEXT: ret void 3524 // 3525 // 3526 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.omp_outlined 3527 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 3528 // CHECK11-NEXT: entry: 3529 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 3530 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 3531 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 3532 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 3533 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3534 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3535 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 3536 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3537 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3538 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 3539 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 3540 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 3541 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3542 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3543 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 3544 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 3545 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 3546 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 3547 // CHECK11-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 3548 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 3549 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3550 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 3551 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3552 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 3553 // CHECK11-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 3554 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3555 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 3556 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3557 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3558 // CHECK11-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 3559 // CHECK11-NEXT: store i32 0, ptr [[I]], align 4 3560 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3561 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 3562 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3563 // CHECK11: omp.precond.then: 3564 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 3565 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3566 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_COMB_UB]], align 4 3567 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 3568 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 3569 // CHECK11-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3570 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4 3571 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP7]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 3572 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3573 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3574 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 3575 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3576 // CHECK11: cond.true: 3577 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3578 // CHECK11-NEXT: br label [[COND_END:%.*]] 3579 // CHECK11: cond.false: 3580 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3581 // CHECK11-NEXT: br label [[COND_END]] 3582 // CHECK11: cond.end: 3583 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 3584 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 3585 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 3586 // CHECK11-NEXT: store i32 [[TMP12]], ptr [[DOTOMP_IV]], align 4 3587 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3588 // CHECK11: omp.inner.for.cond: 3589 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3590 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3591 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 3592 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3593 // CHECK11: omp.inner.for.body: 3594 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 3595 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3596 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, ptr [[N_ADDR]], align 4 3597 // CHECK11-NEXT: store i32 [[TMP17]], ptr [[N_CASTED]], align 4 3598 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, ptr [[N_CASTED]], align 4 3599 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.omp_outlined.omp_outlined, i32 [[TMP15]], i32 [[TMP16]], i32 [[TMP18]], i32 [[TMP0]], ptr [[TMP1]]) 3600 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3601 // CHECK11: omp.inner.for.inc: 3602 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3603 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 3604 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 3605 // CHECK11-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 3606 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 3607 // CHECK11: omp.inner.for.end: 3608 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3609 // CHECK11: omp.loop.exit: 3610 // CHECK11-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3611 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4 3612 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP22]]) 3613 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 3614 // CHECK11: omp.precond.end: 3615 // CHECK11-NEXT: ret void 3616 // 3617 // 3618 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l104.omp_outlined.omp_outlined 3619 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 3620 // CHECK11-NEXT: entry: 3621 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 3622 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 3623 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 3624 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 3625 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 3626 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 3627 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3628 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3629 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 3630 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3631 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3632 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 3633 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3634 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3635 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3636 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3637 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 3638 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 3639 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 3640 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4 3641 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4 3642 // CHECK11-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 3643 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 3644 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3645 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 3646 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3647 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 3648 // CHECK11-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 3649 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3650 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 3651 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3652 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3653 // CHECK11-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 3654 // CHECK11-NEXT: store i32 0, ptr [[I]], align 4 3655 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3656 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 3657 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3658 // CHECK11: omp.precond.then: 3659 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 3660 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3661 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_UB]], align 4 3662 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4 3663 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4 3664 // CHECK11-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_LB]], align 4 3665 // CHECK11-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_UB]], align 4 3666 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 3667 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 3668 // CHECK11-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3669 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4 3670 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP9]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 3671 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3672 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3673 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 3674 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3675 // CHECK11: cond.true: 3676 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3677 // CHECK11-NEXT: br label [[COND_END:%.*]] 3678 // CHECK11: cond.false: 3679 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3680 // CHECK11-NEXT: br label [[COND_END]] 3681 // CHECK11: cond.end: 3682 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 3683 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 3684 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 3685 // CHECK11-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4 3686 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3687 // CHECK11: omp.inner.for.cond: 3688 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3689 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3690 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 3691 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3692 // CHECK11: omp.inner.for.body: 3693 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3694 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 3695 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3696 // CHECK11-NEXT: store i32 [[ADD]], ptr [[I3]], align 4 3697 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, ptr [[I3]], align 4 3698 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 [[TMP18]] 3699 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 3700 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3701 // CHECK11: omp.body.continue: 3702 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3703 // CHECK11: omp.inner.for.inc: 3704 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3705 // CHECK11-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP19]], 1 3706 // CHECK11-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 3707 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 3708 // CHECK11: omp.inner.for.end: 3709 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3710 // CHECK11: omp.loop.exit: 3711 // CHECK11-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3712 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4 3713 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP21]]) 3714 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 3715 // CHECK11: omp.precond.end: 3716 // CHECK11-NEXT: ret void 3717 // 3718 // 3719 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108 3720 // CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 3721 // CHECK11-NEXT: entry: 3722 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 3723 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 3724 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3725 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 3726 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 3727 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 3728 // CHECK11-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 3729 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 3730 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3731 // CHECK11-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 3732 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 3733 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3734 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 3735 // CHECK11-NEXT: store i32 [[TMP2]], ptr [[N_CASTED]], align 4 3736 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_CASTED]], align 4 3737 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 3738 // CHECK11-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 3739 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 3740 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.omp_outlined, i32 [[TMP3]], i32 [[TMP0]], ptr [[TMP1]], i32 [[TMP5]]) 3741 // CHECK11-NEXT: ret void 3742 // 3743 // 3744 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.omp_outlined 3745 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 3746 // CHECK11-NEXT: entry: 3747 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 3748 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 3749 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 3750 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 3751 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3752 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 3753 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3754 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 3755 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3756 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 3757 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 3758 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 3759 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 3760 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3761 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3762 // CHECK11-NEXT: [[I4:%.*]] = alloca i32, align 4 3763 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 3764 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 3765 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 3766 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 3767 // CHECK11-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 3768 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 3769 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3770 // CHECK11-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 3771 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 3772 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3773 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 3774 // CHECK11-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 3775 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3776 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 3777 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3778 // CHECK11-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 3779 // CHECK11-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4 3780 // CHECK11-NEXT: store i32 0, ptr [[I]], align 4 3781 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3782 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 3783 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3784 // CHECK11: omp.precond.then: 3785 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 3786 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 3787 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_COMB_UB]], align 4 3788 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 3789 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 3790 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 3791 // CHECK11-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3792 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 3793 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP6]]) 3794 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3795 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 3796 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 3797 // CHECK11-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3798 // CHECK11: cond.true: 3799 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 3800 // CHECK11-NEXT: br label [[COND_END:%.*]] 3801 // CHECK11: cond.false: 3802 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3803 // CHECK11-NEXT: br label [[COND_END]] 3804 // CHECK11: cond.end: 3805 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 3806 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 3807 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 3808 // CHECK11-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 3809 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3810 // CHECK11: omp.inner.for.cond: 3811 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3812 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 3813 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP15]], 1 3814 // CHECK11-NEXT: [[CMP6:%.*]] = icmp slt i32 [[TMP14]], [[ADD]] 3815 // CHECK11-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3816 // CHECK11: omp.inner.for.body: 3817 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 3818 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3819 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, ptr [[N_ADDR]], align 4 3820 // CHECK11-NEXT: store i32 [[TMP18]], ptr [[N_CASTED]], align 4 3821 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, ptr [[N_CASTED]], align 4 3822 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 3823 // CHECK11-NEXT: store i32 [[TMP20]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 3824 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 3825 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 6, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.omp_outlined.omp_outlined, i32 [[TMP16]], i32 [[TMP17]], i32 [[TMP19]], i32 [[TMP0]], ptr [[TMP1]], i32 [[TMP21]]) 3826 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3827 // CHECK11: omp.inner.for.inc: 3828 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3829 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 3830 // CHECK11-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP22]], [[TMP23]] 3831 // CHECK11-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4 3832 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 3833 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 3834 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP24]], [[TMP25]] 3835 // CHECK11-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_COMB_LB]], align 4 3836 // CHECK11-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3837 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 3838 // CHECK11-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP26]], [[TMP27]] 3839 // CHECK11-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_COMB_UB]], align 4 3840 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3841 // CHECK11-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 3842 // CHECK11-NEXT: [[CMP10:%.*]] = icmp sgt i32 [[TMP28]], [[TMP29]] 3843 // CHECK11-NEXT: br i1 [[CMP10]], label [[COND_TRUE11:%.*]], label [[COND_FALSE12:%.*]] 3844 // CHECK11: cond.true11: 3845 // CHECK11-NEXT: [[TMP30:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 3846 // CHECK11-NEXT: br label [[COND_END13:%.*]] 3847 // CHECK11: cond.false12: 3848 // CHECK11-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3849 // CHECK11-NEXT: br label [[COND_END13]] 3850 // CHECK11: cond.end13: 3851 // CHECK11-NEXT: [[COND14:%.*]] = phi i32 [ [[TMP30]], [[COND_TRUE11]] ], [ [[TMP31]], [[COND_FALSE12]] ] 3852 // CHECK11-NEXT: store i32 [[COND14]], ptr [[DOTOMP_COMB_UB]], align 4 3853 // CHECK11-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 3854 // CHECK11-NEXT: store i32 [[TMP32]], ptr [[DOTOMP_IV]], align 4 3855 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 3856 // CHECK11: omp.inner.for.end: 3857 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3858 // CHECK11: omp.loop.exit: 3859 // CHECK11-NEXT: [[TMP33:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3860 // CHECK11-NEXT: [[TMP34:%.*]] = load i32, ptr [[TMP33]], align 4 3861 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP34]]) 3862 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 3863 // CHECK11: omp.precond.end: 3864 // CHECK11-NEXT: ret void 3865 // 3866 // 3867 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l108.omp_outlined.omp_outlined 3868 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 3869 // CHECK11-NEXT: entry: 3870 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 3871 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 3872 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 3873 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 3874 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 3875 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 3876 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3877 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 3878 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3879 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 3880 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3881 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 3882 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 3883 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3884 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3885 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3886 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3887 // CHECK11-NEXT: [[I4:%.*]] = alloca i32, align 4 3888 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 3889 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 3890 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4 3891 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4 3892 // CHECK11-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 3893 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 3894 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3895 // CHECK11-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 3896 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 3897 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3898 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 3899 // CHECK11-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 3900 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3901 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 3902 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3903 // CHECK11-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 3904 // CHECK11-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4 3905 // CHECK11-NEXT: store i32 0, ptr [[I]], align 4 3906 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3907 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 3908 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3909 // CHECK11: omp.precond.then: 3910 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 3911 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 3912 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_UB]], align 4 3913 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4 3914 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4 3915 // CHECK11-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_LB]], align 4 3916 // CHECK11-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_UB]], align 4 3917 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 3918 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 3919 // CHECK11-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3920 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4 3921 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP9]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 3922 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3923 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 3924 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 3925 // CHECK11-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3926 // CHECK11: cond.true: 3927 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 3928 // CHECK11-NEXT: br label [[COND_END:%.*]] 3929 // CHECK11: cond.false: 3930 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3931 // CHECK11-NEXT: br label [[COND_END]] 3932 // CHECK11: cond.end: 3933 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 3934 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 3935 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 3936 // CHECK11-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4 3937 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3938 // CHECK11: omp.inner.for.cond: 3939 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3940 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3941 // CHECK11-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 3942 // CHECK11-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3943 // CHECK11: omp.inner.for.body: 3944 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3945 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 3946 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3947 // CHECK11-NEXT: store i32 [[ADD]], ptr [[I4]], align 4 3948 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, ptr [[I4]], align 4 3949 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP1]], i32 [[TMP18]] 3950 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 3951 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3952 // CHECK11: omp.body.continue: 3953 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3954 // CHECK11: omp.inner.for.inc: 3955 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3956 // CHECK11-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP19]], 1 3957 // CHECK11-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4 3958 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 3959 // CHECK11: omp.inner.for.end: 3960 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3961 // CHECK11: omp.loop.exit: 3962 // CHECK11-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3963 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4 3964 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP21]]) 3965 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 3966 // CHECK11: omp.precond.end: 3967 // CHECK11-NEXT: ret void 3968 // 3969 // 3970 // CHECK11-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 3971 // CHECK11-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 3972 // CHECK11-NEXT: entry: 3973 // CHECK11-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 3974 // CHECK11-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 3975 // CHECK11-NEXT: [[M:%.*]] = alloca i32, align 4 3976 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 3977 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 3978 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 3979 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 3980 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 3981 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [1 x ptr], align 4 3982 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [1 x ptr], align 4 3983 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [1 x ptr], align 4 3984 // CHECK11-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 3985 // CHECK11-NEXT: [[KERNEL_ARGS5:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 3986 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3987 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 3988 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS8:%.*]] = alloca [2 x ptr], align 4 3989 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS9:%.*]] = alloca [2 x ptr], align 4 3990 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS10:%.*]] = alloca [2 x ptr], align 4 3991 // CHECK11-NEXT: [[_TMP11:%.*]] = alloca i32, align 4 3992 // CHECK11-NEXT: [[KERNEL_ARGS12:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 3993 // CHECK11-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 3994 // CHECK11-NEXT: store i32 10, ptr [[M]], align 4 3995 // CHECK11-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3996 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP0]], align 4 3997 // CHECK11-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3998 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP1]], align 4 3999 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 4000 // CHECK11-NEXT: store ptr null, ptr [[TMP2]], align 4 4001 // CHECK11-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 4002 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 4003 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 4004 // CHECK11-NEXT: store i32 3, ptr [[TMP5]], align 4 4005 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 4006 // CHECK11-NEXT: store i32 1, ptr [[TMP6]], align 4 4007 // CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 4008 // CHECK11-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4 4009 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 4010 // CHECK11-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4 4011 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 4012 // CHECK11-NEXT: store ptr @.offload_sizes.5, ptr [[TMP9]], align 4 4013 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 4014 // CHECK11-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP10]], align 4 4015 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 4016 // CHECK11-NEXT: store ptr null, ptr [[TMP11]], align 4 4017 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 4018 // CHECK11-NEXT: store ptr null, ptr [[TMP12]], align 4 4019 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 4020 // CHECK11-NEXT: store i64 10, ptr [[TMP13]], align 8 4021 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 4022 // CHECK11-NEXT: store i64 0, ptr [[TMP14]], align 8 4023 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 4024 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4 4025 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 4026 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4 4027 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 4028 // CHECK11-NEXT: store i32 0, ptr [[TMP17]], align 4 4029 // CHECK11-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.region_id, ptr [[KERNEL_ARGS]]) 4030 // CHECK11-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0 4031 // CHECK11-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 4032 // CHECK11: omp_offload.failed: 4033 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81(ptr [[A]]) #[[ATTR3]] 4034 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] 4035 // CHECK11: omp_offload.cont: 4036 // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 4037 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP20]], align 4 4038 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 4039 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP21]], align 4 4040 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 4041 // CHECK11-NEXT: store ptr null, ptr [[TMP22]], align 4 4042 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 4043 // CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 4044 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 0 4045 // CHECK11-NEXT: store i32 3, ptr [[TMP25]], align 4 4046 // CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 1 4047 // CHECK11-NEXT: store i32 1, ptr [[TMP26]], align 4 4048 // CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 2 4049 // CHECK11-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 4 4050 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 3 4051 // CHECK11-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 4 4052 // CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 4 4053 // CHECK11-NEXT: store ptr @.offload_sizes.7, ptr [[TMP29]], align 4 4054 // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 5 4055 // CHECK11-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP30]], align 4 4056 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 6 4057 // CHECK11-NEXT: store ptr null, ptr [[TMP31]], align 4 4058 // CHECK11-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 7 4059 // CHECK11-NEXT: store ptr null, ptr [[TMP32]], align 4 4060 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 8 4061 // CHECK11-NEXT: store i64 10, ptr [[TMP33]], align 8 4062 // CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 9 4063 // CHECK11-NEXT: store i64 0, ptr [[TMP34]], align 8 4064 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 10 4065 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4 4066 // CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 11 4067 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4 4068 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 12 4069 // CHECK11-NEXT: store i32 0, ptr [[TMP37]], align 4 4070 // CHECK11-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.region_id, ptr [[KERNEL_ARGS5]]) 4071 // CHECK11-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0 4072 // CHECK11-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]] 4073 // CHECK11: omp_offload.failed6: 4074 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85(ptr [[A]]) #[[ATTR3]] 4075 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT7]] 4076 // CHECK11: omp_offload.cont7: 4077 // CHECK11-NEXT: [[TMP40:%.*]] = load i32, ptr [[M]], align 4 4078 // CHECK11-NEXT: store i32 [[TMP40]], ptr [[DOTCAPTURE_EXPR_]], align 4 4079 // CHECK11-NEXT: [[TMP41:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 4080 // CHECK11-NEXT: store i32 [[TMP41]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 4081 // CHECK11-NEXT: [[TMP42:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 4082 // CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0 4083 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP43]], align 4 4084 // CHECK11-NEXT: [[TMP44:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0 4085 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP44]], align 4 4086 // CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i32 0, i32 0 4087 // CHECK11-NEXT: store ptr null, ptr [[TMP45]], align 4 4088 // CHECK11-NEXT: [[TMP46:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 1 4089 // CHECK11-NEXT: store i32 [[TMP42]], ptr [[TMP46]], align 4 4090 // CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 1 4091 // CHECK11-NEXT: store i32 [[TMP42]], ptr [[TMP47]], align 4 4092 // CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i32 0, i32 1 4093 // CHECK11-NEXT: store ptr null, ptr [[TMP48]], align 4 4094 // CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0 4095 // CHECK11-NEXT: [[TMP50:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0 4096 // CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 0 4097 // CHECK11-NEXT: store i32 3, ptr [[TMP51]], align 4 4098 // CHECK11-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 1 4099 // CHECK11-NEXT: store i32 2, ptr [[TMP52]], align 4 4100 // CHECK11-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 2 4101 // CHECK11-NEXT: store ptr [[TMP49]], ptr [[TMP53]], align 4 4102 // CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 3 4103 // CHECK11-NEXT: store ptr [[TMP50]], ptr [[TMP54]], align 4 4104 // CHECK11-NEXT: [[TMP55:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 4 4105 // CHECK11-NEXT: store ptr @.offload_sizes.9, ptr [[TMP55]], align 4 4106 // CHECK11-NEXT: [[TMP56:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 5 4107 // CHECK11-NEXT: store ptr @.offload_maptypes.10, ptr [[TMP56]], align 4 4108 // CHECK11-NEXT: [[TMP57:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 6 4109 // CHECK11-NEXT: store ptr null, ptr [[TMP57]], align 4 4110 // CHECK11-NEXT: [[TMP58:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 7 4111 // CHECK11-NEXT: store ptr null, ptr [[TMP58]], align 4 4112 // CHECK11-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 8 4113 // CHECK11-NEXT: store i64 10, ptr [[TMP59]], align 8 4114 // CHECK11-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 9 4115 // CHECK11-NEXT: store i64 0, ptr [[TMP60]], align 8 4116 // CHECK11-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 10 4117 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP61]], align 4 4118 // CHECK11-NEXT: [[TMP62:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 11 4119 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP62]], align 4 4120 // CHECK11-NEXT: [[TMP63:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 12 4121 // CHECK11-NEXT: store i32 0, ptr [[TMP63]], align 4 4122 // CHECK11-NEXT: [[TMP64:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, ptr [[KERNEL_ARGS12]]) 4123 // CHECK11-NEXT: [[TMP65:%.*]] = icmp ne i32 [[TMP64]], 0 4124 // CHECK11-NEXT: br i1 [[TMP65]], label [[OMP_OFFLOAD_FAILED13:%.*]], label [[OMP_OFFLOAD_CONT14:%.*]] 4125 // CHECK11: omp_offload.failed13: 4126 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89(ptr [[A]], i32 [[TMP42]]) #[[ATTR3]] 4127 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT14]] 4128 // CHECK11: omp_offload.cont14: 4129 // CHECK11-NEXT: ret i32 0 4130 // 4131 // 4132 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81 4133 // CHECK11-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 4134 // CHECK11-NEXT: entry: 4135 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 4136 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 4137 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 4138 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.omp_outlined, ptr [[TMP0]]) 4139 // CHECK11-NEXT: ret void 4140 // 4141 // 4142 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.omp_outlined 4143 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 4144 // CHECK11-NEXT: entry: 4145 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 4146 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 4147 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 4148 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4149 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 4150 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 4151 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 4152 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 4153 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 4154 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 4155 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 4156 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 4157 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 4158 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 4159 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 4160 // CHECK11-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4 4161 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 4162 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 4163 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 4164 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 4165 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 4166 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 4167 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 4168 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 4169 // CHECK11: cond.true: 4170 // CHECK11-NEXT: br label [[COND_END:%.*]] 4171 // CHECK11: cond.false: 4172 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 4173 // CHECK11-NEXT: br label [[COND_END]] 4174 // CHECK11: cond.end: 4175 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 4176 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 4177 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 4178 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 4179 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4180 // CHECK11: omp.inner.for.cond: 4181 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4182 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 4183 // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 4184 // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4185 // CHECK11: omp.inner.for.body: 4186 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 4187 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 4188 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]]) 4189 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4190 // CHECK11: omp.inner.for.inc: 4191 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4192 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 4193 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] 4194 // CHECK11-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 4195 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 4196 // CHECK11: omp.inner.for.end: 4197 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 4198 // CHECK11: omp.loop.exit: 4199 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 4200 // CHECK11-NEXT: ret void 4201 // 4202 // 4203 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l81.omp_outlined.omp_outlined 4204 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 4205 // CHECK11-NEXT: entry: 4206 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 4207 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 4208 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 4209 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 4210 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 4211 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4212 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 4213 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4214 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4215 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 4216 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 4217 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 4218 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 4219 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 4220 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4 4221 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4 4222 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 4223 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 4224 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 4225 // CHECK11-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 4226 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4 4227 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4 4228 // CHECK11-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4 4229 // CHECK11-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4 4230 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 4231 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 4232 // CHECK11-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 4233 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 4234 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 4235 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4236 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9 4237 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 4238 // CHECK11: cond.true: 4239 // CHECK11-NEXT: br label [[COND_END:%.*]] 4240 // CHECK11: cond.false: 4241 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4242 // CHECK11-NEXT: br label [[COND_END]] 4243 // CHECK11: cond.end: 4244 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 4245 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 4246 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 4247 // CHECK11-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 4248 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4249 // CHECK11: omp.inner.for.cond: 4250 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4251 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4252 // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 4253 // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4254 // CHECK11: omp.inner.for.body: 4255 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4256 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 4257 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4258 // CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4 4259 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4 4260 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP11]] 4261 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 4262 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4263 // CHECK11: omp.body.continue: 4264 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4265 // CHECK11: omp.inner.for.inc: 4266 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4267 // CHECK11-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1 4268 // CHECK11-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 4269 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 4270 // CHECK11: omp.inner.for.end: 4271 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 4272 // CHECK11: omp.loop.exit: 4273 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]]) 4274 // CHECK11-NEXT: ret void 4275 // 4276 // 4277 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85 4278 // CHECK11-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 4279 // CHECK11-NEXT: entry: 4280 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 4281 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 4282 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 4283 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.omp_outlined, ptr [[TMP0]]) 4284 // CHECK11-NEXT: ret void 4285 // 4286 // 4287 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.omp_outlined 4288 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 4289 // CHECK11-NEXT: entry: 4290 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 4291 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 4292 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 4293 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4294 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 4295 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 4296 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 4297 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 4298 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 4299 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 4300 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 4301 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 4302 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 4303 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 4304 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 4305 // CHECK11-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4 4306 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 4307 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 4308 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 4309 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 4310 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 4311 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 4312 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 4313 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 4314 // CHECK11: cond.true: 4315 // CHECK11-NEXT: br label [[COND_END:%.*]] 4316 // CHECK11: cond.false: 4317 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 4318 // CHECK11-NEXT: br label [[COND_END]] 4319 // CHECK11: cond.end: 4320 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 4321 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 4322 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 4323 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 4324 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4325 // CHECK11: omp.inner.for.cond: 4326 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4327 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 4328 // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 4329 // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4330 // CHECK11: omp.inner.for.body: 4331 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 4332 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 4333 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]]) 4334 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4335 // CHECK11: omp.inner.for.inc: 4336 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4337 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 4338 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] 4339 // CHECK11-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 4340 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 4341 // CHECK11: omp.inner.for.end: 4342 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 4343 // CHECK11: omp.loop.exit: 4344 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 4345 // CHECK11-NEXT: ret void 4346 // 4347 // 4348 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l85.omp_outlined.omp_outlined 4349 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 4350 // CHECK11-NEXT: entry: 4351 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 4352 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 4353 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 4354 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 4355 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 4356 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4357 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 4358 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4359 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4360 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 4361 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 4362 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 4363 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 4364 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 4365 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4 4366 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4 4367 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 4368 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 4369 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 4370 // CHECK11-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 4371 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4 4372 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4 4373 // CHECK11-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4 4374 // CHECK11-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4 4375 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 4376 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 4377 // CHECK11-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 4378 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 4379 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 4380 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4381 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9 4382 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 4383 // CHECK11: cond.true: 4384 // CHECK11-NEXT: br label [[COND_END:%.*]] 4385 // CHECK11: cond.false: 4386 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4387 // CHECK11-NEXT: br label [[COND_END]] 4388 // CHECK11: cond.end: 4389 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 4390 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 4391 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 4392 // CHECK11-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 4393 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4394 // CHECK11: omp.inner.for.cond: 4395 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4396 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4397 // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 4398 // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4399 // CHECK11: omp.inner.for.body: 4400 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4401 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 4402 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4403 // CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4 4404 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4 4405 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP11]] 4406 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 4407 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4408 // CHECK11: omp.body.continue: 4409 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4410 // CHECK11: omp.inner.for.inc: 4411 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4412 // CHECK11-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1 4413 // CHECK11-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 4414 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 4415 // CHECK11: omp.inner.for.end: 4416 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 4417 // CHECK11: omp.loop.exit: 4418 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]]) 4419 // CHECK11-NEXT: ret void 4420 // 4421 // 4422 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89 4423 // CHECK11-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 4424 // CHECK11-NEXT: entry: 4425 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 4426 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 4427 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 4428 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 4429 // CHECK11-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 4430 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 4431 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 4432 // CHECK11-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 4433 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 4434 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined, ptr [[TMP0]], i32 [[TMP2]]) 4435 // CHECK11-NEXT: ret void 4436 // 4437 // 4438 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined 4439 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 4440 // CHECK11-NEXT: entry: 4441 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 4442 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 4443 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 4444 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 4445 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4446 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 4447 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 4448 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 4449 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 4450 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 4451 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 4452 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 4453 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 4454 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 4455 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 4456 // CHECK11-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 4457 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 4458 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 4459 // CHECK11-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4 4460 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 4461 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 4462 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 4463 // CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 4464 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4 4465 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP3]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP1]]) 4466 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 4467 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 9 4468 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 4469 // CHECK11: cond.true: 4470 // CHECK11-NEXT: br label [[COND_END:%.*]] 4471 // CHECK11: cond.false: 4472 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 4473 // CHECK11-NEXT: br label [[COND_END]] 4474 // CHECK11: cond.end: 4475 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ] 4476 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 4477 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 4478 // CHECK11-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4 4479 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4480 // CHECK11: omp.inner.for.cond: 4481 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4482 // CHECK11-NEXT: [[CMP1:%.*]] = icmp slt i32 [[TMP7]], 10 4483 // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4484 // CHECK11: omp.inner.for.body: 4485 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 4486 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 4487 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 4488 // CHECK11-NEXT: store i32 [[TMP10]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 4489 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 4490 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]], i32 [[TMP11]]) 4491 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4492 // CHECK11: omp.inner.for.inc: 4493 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4494 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 4495 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] 4496 // CHECK11-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 4497 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 4498 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 4499 // CHECK11-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP14]], [[TMP15]] 4500 // CHECK11-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_COMB_LB]], align 4 4501 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 4502 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 4503 // CHECK11-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP16]], [[TMP17]] 4504 // CHECK11-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_COMB_UB]], align 4 4505 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 4506 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP18]], 9 4507 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE5:%.*]], label [[COND_FALSE6:%.*]] 4508 // CHECK11: cond.true5: 4509 // CHECK11-NEXT: br label [[COND_END7:%.*]] 4510 // CHECK11: cond.false6: 4511 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 4512 // CHECK11-NEXT: br label [[COND_END7]] 4513 // CHECK11: cond.end7: 4514 // CHECK11-NEXT: [[COND8:%.*]] = phi i32 [ 9, [[COND_TRUE5]] ], [ [[TMP19]], [[COND_FALSE6]] ] 4515 // CHECK11-NEXT: store i32 [[COND8]], ptr [[DOTOMP_COMB_UB]], align 4 4516 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 4517 // CHECK11-NEXT: store i32 [[TMP20]], ptr [[DOTOMP_IV]], align 4 4518 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 4519 // CHECK11: omp.inner.for.end: 4520 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 4521 // CHECK11: omp.loop.exit: 4522 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP3]]) 4523 // CHECK11-NEXT: ret void 4524 // 4525 // 4526 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined.omp_outlined 4527 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 4528 // CHECK11-NEXT: entry: 4529 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 4530 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 4531 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 4532 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 4533 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 4534 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 4535 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4536 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 4537 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4538 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4539 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 4540 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 4541 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 4542 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 4543 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 4544 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4 4545 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4 4546 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 4547 // CHECK11-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 4548 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 4549 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 4550 // CHECK11-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 4551 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4 4552 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4 4553 // CHECK11-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4 4554 // CHECK11-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4 4555 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 4556 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 4557 // CHECK11-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 4558 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 4559 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 4560 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4561 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9 4562 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 4563 // CHECK11: cond.true: 4564 // CHECK11-NEXT: br label [[COND_END:%.*]] 4565 // CHECK11: cond.false: 4566 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4567 // CHECK11-NEXT: br label [[COND_END]] 4568 // CHECK11: cond.end: 4569 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 4570 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 4571 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 4572 // CHECK11-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 4573 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4574 // CHECK11: omp.inner.for.cond: 4575 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4576 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4577 // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 4578 // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4579 // CHECK11: omp.inner.for.body: 4580 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4581 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 4582 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4583 // CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4 4584 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4 4585 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP11]] 4586 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 4587 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4588 // CHECK11: omp.body.continue: 4589 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4590 // CHECK11: omp.inner.for.inc: 4591 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4592 // CHECK11-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1 4593 // CHECK11-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 4594 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 4595 // CHECK11: omp.inner.for.end: 4596 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 4597 // CHECK11: omp.loop.exit: 4598 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]]) 4599 // CHECK11-NEXT: ret void 4600 // 4601