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 29 #pragma omp teams distribute 30 for(int i = 0; i < X; i++) { 31 a[i] = (T)0; 32 } 33 #pragma omp target 34 #pragma omp teams distribute dist_schedule(static) 35 for(int i = 0; i < X; i++) { 36 a[i] = (T)0; 37 } 38 #pragma omp target 39 #pragma omp teams distribute dist_schedule(static, X/2) 40 for(int i = 0; i < X; i++) { 41 a[i] = (T)0; 42 } 43 44 45 46 47 48 49 return a[0]; 50 } 51 }; 52 53 int teams_template_struct(void) { 54 SS<int, 123, 456> V; 55 return V.foo(); 56 57 } 58 #endif // CK1 59 60 // Test host codegen. 61 // 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 62 // 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 63 // 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 64 // 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 65 // 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 66 // 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 67 68 // 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}}" 69 // 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 70 // 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}}" 71 // 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}}" 72 // 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 73 // 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}}" 74 #ifdef CK2 75 76 template <typename T, int n> 77 int tmain(T argc) { 78 T a[n]; 79 #pragma omp target 80 #pragma omp teams distribute 81 for(int i = 0; i < n; i++) { 82 a[i] = (T)0; 83 } 84 #pragma omp target 85 #pragma omp teams distribute dist_schedule(static) 86 for(int i = 0; i < n; i++) { 87 a[i] = (T)0; 88 } 89 #pragma omp target 90 #pragma omp teams distribute dist_schedule(static, n) 91 for(int i = 0; i < n; i++) { 92 a[i] = (T)0; 93 } 94 return 0; 95 } 96 97 int main (int argc, char **argv) { 98 int n = 100; 99 int a[n]; 100 #pragma omp target 101 #pragma omp teams distribute 102 for(int i = 0; i < n; i++) { 103 a[i] = 0; 104 } 105 #pragma omp target 106 #pragma omp teams distribute dist_schedule(static) 107 for(int i = 0; i < n; i++) { 108 a[i] = 0; 109 } 110 #pragma omp target 111 #pragma omp teams distribute dist_schedule(static, n) 112 for(int i = 0; i < n; i++) { 113 a[i] = 0; 114 } 115 return tmain<int, 10>(argc); 116 } 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 #endif // CK2 133 #endif // #ifndef HEADER 134 // CHECK1-LABEL: define {{[^@]+}}@_Z21teams_template_structv 135 // CHECK1-SAME: () #[[ATTR0:[0-9]+]] { 136 // CHECK1-NEXT: entry: 137 // CHECK1-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 138 // CHECK1-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]]) 139 // CHECK1-NEXT: ret i32 [[CALL]] 140 // 141 // 142 // CHECK1-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 143 // CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat { 144 // CHECK1-NEXT: entry: 145 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 146 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 147 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 148 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 149 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 150 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 151 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x ptr], align 8 152 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x ptr], align 8 153 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x ptr], align 8 154 // CHECK1-NEXT: [[_TMP6:%.*]] = alloca i32, align 4 155 // CHECK1-NEXT: [[KERNEL_ARGS7:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 156 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS11:%.*]] = alloca [1 x ptr], align 8 157 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS12:%.*]] = alloca [1 x ptr], align 8 158 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS13:%.*]] = alloca [1 x ptr], align 8 159 // CHECK1-NEXT: [[_TMP14:%.*]] = alloca i32, align 4 160 // CHECK1-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 161 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 162 // CHECK1-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 163 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 164 // CHECK1-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 165 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 8 166 // CHECK1-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 167 // CHECK1-NEXT: store ptr [[A]], ptr [[TMP1]], align 8 168 // CHECK1-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 169 // CHECK1-NEXT: store ptr null, ptr [[TMP2]], align 8 170 // CHECK1-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 171 // CHECK1-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 172 // CHECK1-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 173 // CHECK1-NEXT: store i32 3, ptr [[TMP5]], align 4 174 // CHECK1-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 175 // CHECK1-NEXT: store i32 1, ptr [[TMP6]], align 4 176 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 177 // CHECK1-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8 178 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 179 // CHECK1-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8 180 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 181 // CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 8 182 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 183 // CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 8 184 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 185 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8 186 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 187 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8 188 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 189 // CHECK1-NEXT: store i64 123, ptr [[TMP13]], align 8 190 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 191 // CHECK1-NEXT: store i64 0, ptr [[TMP14]], align 8 192 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 193 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4 194 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 195 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4 196 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 197 // CHECK1-NEXT: store i32 0, ptr [[TMP17]], align 4 198 // CHECK1-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]]) 199 // CHECK1-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0 200 // CHECK1-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 201 // CHECK1: omp_offload.failed: 202 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]] 203 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] 204 // CHECK1: omp_offload.cont: 205 // CHECK1-NEXT: [[A2:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 206 // CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 207 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP20]], align 8 208 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 209 // CHECK1-NEXT: store ptr [[A2]], ptr [[TMP21]], align 8 210 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS5]], i64 0, i64 0 211 // CHECK1-NEXT: store ptr null, ptr [[TMP22]], align 8 212 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 213 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 214 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 0 215 // CHECK1-NEXT: store i32 3, ptr [[TMP25]], align 4 216 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 1 217 // CHECK1-NEXT: store i32 1, ptr [[TMP26]], align 4 218 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 2 219 // CHECK1-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 8 220 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 3 221 // CHECK1-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 8 222 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 4 223 // CHECK1-NEXT: store ptr @.offload_sizes.1, ptr [[TMP29]], align 8 224 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 5 225 // CHECK1-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP30]], align 8 226 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 6 227 // CHECK1-NEXT: store ptr null, ptr [[TMP31]], align 8 228 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 7 229 // CHECK1-NEXT: store ptr null, ptr [[TMP32]], align 8 230 // CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 8 231 // CHECK1-NEXT: store i64 123, ptr [[TMP33]], align 8 232 // CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 9 233 // CHECK1-NEXT: store i64 0, ptr [[TMP34]], align 8 234 // CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 10 235 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4 236 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 11 237 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4 238 // CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 12 239 // CHECK1-NEXT: store i32 0, ptr [[TMP37]], align 4 240 // CHECK1-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33.region_id, ptr [[KERNEL_ARGS7]]) 241 // CHECK1-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0 242 // CHECK1-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED8:%.*]], label [[OMP_OFFLOAD_CONT9:%.*]] 243 // CHECK1: omp_offload.failed8: 244 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33(ptr [[THIS1]]) #[[ATTR2]] 245 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT9]] 246 // CHECK1: omp_offload.cont9: 247 // CHECK1-NEXT: [[A10:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 248 // CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0 249 // CHECK1-NEXT: store ptr [[THIS1]], ptr [[TMP40]], align 8 250 // CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0 251 // CHECK1-NEXT: store ptr [[A10]], ptr [[TMP41]], align 8 252 // CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS13]], i64 0, i64 0 253 // CHECK1-NEXT: store ptr null, ptr [[TMP42]], align 8 254 // CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0 255 // CHECK1-NEXT: [[TMP44:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0 256 // CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0 257 // CHECK1-NEXT: store i32 3, ptr [[TMP45]], align 4 258 // CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1 259 // CHECK1-NEXT: store i32 1, ptr [[TMP46]], align 4 260 // CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2 261 // CHECK1-NEXT: store ptr [[TMP43]], ptr [[TMP47]], align 8 262 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3 263 // CHECK1-NEXT: store ptr [[TMP44]], ptr [[TMP48]], align 8 264 // CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4 265 // CHECK1-NEXT: store ptr @.offload_sizes.3, ptr [[TMP49]], align 8 266 // CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5 267 // CHECK1-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP50]], align 8 268 // CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6 269 // CHECK1-NEXT: store ptr null, ptr [[TMP51]], align 8 270 // CHECK1-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7 271 // CHECK1-NEXT: store ptr null, ptr [[TMP52]], align 8 272 // CHECK1-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8 273 // CHECK1-NEXT: store i64 123, ptr [[TMP53]], align 8 274 // CHECK1-NEXT: [[TMP54:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9 275 // CHECK1-NEXT: store i64 0, ptr [[TMP54]], align 8 276 // CHECK1-NEXT: [[TMP55:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10 277 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP55]], align 4 278 // CHECK1-NEXT: [[TMP56:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11 279 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP56]], align 4 280 // CHECK1-NEXT: [[TMP57:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12 281 // CHECK1-NEXT: store i32 0, ptr [[TMP57]], align 4 282 // CHECK1-NEXT: [[TMP58:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38.region_id, ptr [[KERNEL_ARGS15]]) 283 // CHECK1-NEXT: [[TMP59:%.*]] = icmp ne i32 [[TMP58]], 0 284 // CHECK1-NEXT: br i1 [[TMP59]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] 285 // CHECK1: omp_offload.failed16: 286 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38(ptr [[THIS1]]) #[[ATTR2]] 287 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT17]] 288 // CHECK1: omp_offload.cont17: 289 // CHECK1-NEXT: [[A18:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 290 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A18]], i64 0, i64 0 291 // CHECK1-NEXT: [[TMP60:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 292 // CHECK1-NEXT: ret i32 [[TMP60]] 293 // 294 // 295 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28 296 // CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 297 // CHECK1-NEXT: entry: 298 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 299 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 300 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 301 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined, ptr [[TMP0]]) 302 // CHECK1-NEXT: ret void 303 // 304 // 305 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined 306 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 307 // CHECK1-NEXT: entry: 308 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 309 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 310 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 311 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 312 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 313 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 314 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 315 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 316 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 317 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 318 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 319 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 320 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 321 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 322 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 323 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 324 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 325 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 326 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 327 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 328 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 329 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 330 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 331 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 332 // CHECK1: cond.true: 333 // CHECK1-NEXT: br label [[COND_END:%.*]] 334 // CHECK1: cond.false: 335 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 336 // CHECK1-NEXT: br label [[COND_END]] 337 // CHECK1: cond.end: 338 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 339 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 340 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 341 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 342 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 343 // CHECK1: omp.inner.for.cond: 344 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 345 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 346 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 347 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 348 // CHECK1: omp.inner.for.body: 349 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 350 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 351 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 352 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4 353 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 354 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4 355 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP9]] to i64 356 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]] 357 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 358 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 359 // CHECK1: omp.body.continue: 360 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 361 // CHECK1: omp.inner.for.inc: 362 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 363 // CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 364 // CHECK1-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 365 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 366 // CHECK1: omp.inner.for.end: 367 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 368 // CHECK1: omp.loop.exit: 369 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 370 // CHECK1-NEXT: ret void 371 // 372 // 373 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33 374 // CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 375 // CHECK1-NEXT: entry: 376 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 377 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 378 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 379 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33.omp_outlined, ptr [[TMP0]]) 380 // CHECK1-NEXT: ret void 381 // 382 // 383 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33.omp_outlined 384 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 385 // CHECK1-NEXT: entry: 386 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 387 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 388 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 389 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 390 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 391 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 392 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 393 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 394 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 395 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 396 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 397 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 398 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 399 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 400 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 401 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 402 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 403 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 404 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 405 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 406 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 407 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 408 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 409 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 410 // CHECK1: cond.true: 411 // CHECK1-NEXT: br label [[COND_END:%.*]] 412 // CHECK1: cond.false: 413 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 414 // CHECK1-NEXT: br label [[COND_END]] 415 // CHECK1: cond.end: 416 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 417 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 418 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 419 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 420 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 421 // CHECK1: omp.inner.for.cond: 422 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 423 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 424 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 425 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 426 // CHECK1: omp.inner.for.body: 427 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 428 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 429 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 430 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4 431 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 432 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4 433 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP9]] to i64 434 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]] 435 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 436 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 437 // CHECK1: omp.body.continue: 438 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 439 // CHECK1: omp.inner.for.inc: 440 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 441 // CHECK1-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 442 // CHECK1-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 443 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 444 // CHECK1: omp.inner.for.end: 445 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 446 // CHECK1: omp.loop.exit: 447 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 448 // CHECK1-NEXT: ret void 449 // 450 // 451 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38 452 // CHECK1-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 453 // CHECK1-NEXT: entry: 454 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 455 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 456 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 457 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38.omp_outlined, ptr [[TMP0]]) 458 // CHECK1-NEXT: ret void 459 // 460 // 461 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38.omp_outlined 462 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 463 // CHECK1-NEXT: entry: 464 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 465 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 466 // CHECK1-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 467 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 468 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 469 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 470 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 471 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 472 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 473 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 474 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 475 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 476 // CHECK1-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 477 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 478 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 479 // CHECK1-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 480 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 481 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 482 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 483 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 484 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 61) 485 // CHECK1-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 486 // CHECK1: omp.dispatch.cond: 487 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 488 // CHECK1-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 489 // CHECK1-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 490 // CHECK1: cond.true: 491 // CHECK1-NEXT: br label [[COND_END:%.*]] 492 // CHECK1: cond.false: 493 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 494 // CHECK1-NEXT: br label [[COND_END]] 495 // CHECK1: cond.end: 496 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 497 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 498 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 499 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 500 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 501 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 502 // CHECK1-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 503 // CHECK1-NEXT: br i1 [[CMP1]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 504 // CHECK1: omp.dispatch.body: 505 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 506 // CHECK1: omp.inner.for.cond: 507 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8:![0-9]+]] 508 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP8]] 509 // CHECK1-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 510 // CHECK1-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 511 // CHECK1: omp.inner.for.body: 512 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8]] 513 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 514 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 515 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP8]] 516 // CHECK1-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 517 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP8]] 518 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 519 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]] 520 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP8]] 521 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 522 // CHECK1: omp.body.continue: 523 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 524 // CHECK1: omp.inner.for.inc: 525 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8]] 526 // CHECK1-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1 527 // CHECK1-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8]] 528 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]] 529 // CHECK1: omp.inner.for.end: 530 // CHECK1-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 531 // CHECK1: omp.dispatch.inc: 532 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 533 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 534 // CHECK1-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP13]], [[TMP14]] 535 // CHECK1-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_LB]], align 4 536 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 537 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 538 // CHECK1-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP15]], [[TMP16]] 539 // CHECK1-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4 540 // CHECK1-NEXT: br label [[OMP_DISPATCH_COND]] 541 // CHECK1: omp.dispatch.end: 542 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 543 // CHECK1-NEXT: ret void 544 // 545 // 546 // CHECK3-LABEL: define {{[^@]+}}@_Z21teams_template_structv 547 // CHECK3-SAME: () #[[ATTR0:[0-9]+]] { 548 // CHECK3-NEXT: entry: 549 // CHECK3-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 550 // CHECK3-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]]) 551 // CHECK3-NEXT: ret i32 [[CALL]] 552 // 553 // 554 // CHECK3-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 555 // CHECK3-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 556 // CHECK3-NEXT: entry: 557 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 558 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 559 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 560 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 561 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 562 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 563 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS3:%.*]] = alloca [1 x ptr], align 4 564 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS4:%.*]] = alloca [1 x ptr], align 4 565 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS5:%.*]] = alloca [1 x ptr], align 4 566 // CHECK3-NEXT: [[_TMP6:%.*]] = alloca i32, align 4 567 // CHECK3-NEXT: [[KERNEL_ARGS7:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 568 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS11:%.*]] = alloca [1 x ptr], align 4 569 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS12:%.*]] = alloca [1 x ptr], align 4 570 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS13:%.*]] = alloca [1 x ptr], align 4 571 // CHECK3-NEXT: [[_TMP14:%.*]] = alloca i32, align 4 572 // CHECK3-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 573 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 574 // CHECK3-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 575 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 576 // CHECK3-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 577 // CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 4 578 // CHECK3-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 579 // CHECK3-NEXT: store ptr [[A]], ptr [[TMP1]], align 4 580 // CHECK3-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 581 // CHECK3-NEXT: store ptr null, ptr [[TMP2]], align 4 582 // CHECK3-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 583 // CHECK3-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 584 // CHECK3-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 585 // CHECK3-NEXT: store i32 3, ptr [[TMP5]], align 4 586 // CHECK3-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 587 // CHECK3-NEXT: store i32 1, ptr [[TMP6]], align 4 588 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 589 // CHECK3-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4 590 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 591 // CHECK3-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4 592 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 593 // CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 4 594 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 595 // CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 4 596 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 597 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4 598 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 599 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4 600 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 601 // CHECK3-NEXT: store i64 123, ptr [[TMP13]], align 8 602 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 603 // CHECK3-NEXT: store i64 0, ptr [[TMP14]], align 8 604 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 605 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4 606 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 607 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4 608 // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 609 // CHECK3-NEXT: store i32 0, ptr [[TMP17]], align 4 610 // CHECK3-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.region_id, ptr [[KERNEL_ARGS]]) 611 // CHECK3-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0 612 // CHECK3-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 613 // CHECK3: omp_offload.failed: 614 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28(ptr [[THIS1]]) #[[ATTR2:[0-9]+]] 615 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] 616 // CHECK3: omp_offload.cont: 617 // CHECK3-NEXT: [[A2:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 618 // CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 619 // CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP20]], align 4 620 // CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 621 // CHECK3-NEXT: store ptr [[A2]], ptr [[TMP21]], align 4 622 // CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS5]], i32 0, i32 0 623 // CHECK3-NEXT: store ptr null, ptr [[TMP22]], align 4 624 // CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS3]], i32 0, i32 0 625 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS4]], i32 0, i32 0 626 // CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 0 627 // CHECK3-NEXT: store i32 3, ptr [[TMP25]], align 4 628 // CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 1 629 // CHECK3-NEXT: store i32 1, ptr [[TMP26]], align 4 630 // CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 2 631 // CHECK3-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 4 632 // CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 3 633 // CHECK3-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 4 634 // CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 4 635 // CHECK3-NEXT: store ptr @.offload_sizes.1, ptr [[TMP29]], align 4 636 // CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 5 637 // CHECK3-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP30]], align 4 638 // CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 6 639 // CHECK3-NEXT: store ptr null, ptr [[TMP31]], align 4 640 // CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 7 641 // CHECK3-NEXT: store ptr null, ptr [[TMP32]], align 4 642 // CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 8 643 // CHECK3-NEXT: store i64 123, ptr [[TMP33]], align 8 644 // CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 9 645 // CHECK3-NEXT: store i64 0, ptr [[TMP34]], align 8 646 // CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 10 647 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4 648 // CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 11 649 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4 650 // CHECK3-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS7]], i32 0, i32 12 651 // CHECK3-NEXT: store i32 0, ptr [[TMP37]], align 4 652 // CHECK3-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33.region_id, ptr [[KERNEL_ARGS7]]) 653 // CHECK3-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0 654 // CHECK3-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED8:%.*]], label [[OMP_OFFLOAD_CONT9:%.*]] 655 // CHECK3: omp_offload.failed8: 656 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33(ptr [[THIS1]]) #[[ATTR2]] 657 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT9]] 658 // CHECK3: omp_offload.cont9: 659 // CHECK3-NEXT: [[A10:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 660 // CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0 661 // CHECK3-NEXT: store ptr [[THIS1]], ptr [[TMP40]], align 4 662 // CHECK3-NEXT: [[TMP41:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0 663 // CHECK3-NEXT: store ptr [[A10]], ptr [[TMP41]], align 4 664 // CHECK3-NEXT: [[TMP42:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS13]], i32 0, i32 0 665 // CHECK3-NEXT: store ptr null, ptr [[TMP42]], align 4 666 // CHECK3-NEXT: [[TMP43:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS11]], i32 0, i32 0 667 // CHECK3-NEXT: [[TMP44:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS12]], i32 0, i32 0 668 // CHECK3-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0 669 // CHECK3-NEXT: store i32 3, ptr [[TMP45]], align 4 670 // CHECK3-NEXT: [[TMP46:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1 671 // CHECK3-NEXT: store i32 1, ptr [[TMP46]], align 4 672 // CHECK3-NEXT: [[TMP47:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2 673 // CHECK3-NEXT: store ptr [[TMP43]], ptr [[TMP47]], align 4 674 // CHECK3-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3 675 // CHECK3-NEXT: store ptr [[TMP44]], ptr [[TMP48]], align 4 676 // CHECK3-NEXT: [[TMP49:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4 677 // CHECK3-NEXT: store ptr @.offload_sizes.3, ptr [[TMP49]], align 4 678 // CHECK3-NEXT: [[TMP50:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5 679 // CHECK3-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP50]], align 4 680 // CHECK3-NEXT: [[TMP51:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6 681 // CHECK3-NEXT: store ptr null, ptr [[TMP51]], align 4 682 // CHECK3-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7 683 // CHECK3-NEXT: store ptr null, ptr [[TMP52]], align 4 684 // CHECK3-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8 685 // CHECK3-NEXT: store i64 123, ptr [[TMP53]], align 8 686 // CHECK3-NEXT: [[TMP54:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9 687 // CHECK3-NEXT: store i64 0, ptr [[TMP54]], align 8 688 // CHECK3-NEXT: [[TMP55:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10 689 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP55]], align 4 690 // CHECK3-NEXT: [[TMP56:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11 691 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP56]], align 4 692 // CHECK3-NEXT: [[TMP57:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12 693 // CHECK3-NEXT: store i32 0, ptr [[TMP57]], align 4 694 // CHECK3-NEXT: [[TMP58:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38.region_id, ptr [[KERNEL_ARGS15]]) 695 // CHECK3-NEXT: [[TMP59:%.*]] = icmp ne i32 [[TMP58]], 0 696 // CHECK3-NEXT: br i1 [[TMP59]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] 697 // CHECK3: omp_offload.failed16: 698 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38(ptr [[THIS1]]) #[[ATTR2]] 699 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT17]] 700 // CHECK3: omp_offload.cont17: 701 // CHECK3-NEXT: [[A18:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 702 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A18]], i32 0, i32 0 703 // CHECK3-NEXT: [[TMP60:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 704 // CHECK3-NEXT: ret i32 [[TMP60]] 705 // 706 // 707 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28 708 // CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 709 // CHECK3-NEXT: entry: 710 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 711 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 712 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 713 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined, ptr [[TMP0]]) 714 // CHECK3-NEXT: ret void 715 // 716 // 717 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l28.omp_outlined 718 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 719 // CHECK3-NEXT: entry: 720 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 721 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 722 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 723 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 724 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 725 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 726 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 727 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 728 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 729 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 730 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 731 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 732 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 733 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 734 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 735 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 736 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 737 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 738 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 739 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 740 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 741 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 742 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 743 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 744 // CHECK3: cond.true: 745 // CHECK3-NEXT: br label [[COND_END:%.*]] 746 // CHECK3: cond.false: 747 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 748 // CHECK3-NEXT: br label [[COND_END]] 749 // CHECK3: cond.end: 750 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 751 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 752 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 753 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 754 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 755 // CHECK3: omp.inner.for.cond: 756 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 757 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 758 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 759 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 760 // CHECK3: omp.inner.for.body: 761 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 762 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 763 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 764 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4 765 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 766 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4 767 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP9]] 768 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 769 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 770 // CHECK3: omp.body.continue: 771 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 772 // CHECK3: omp.inner.for.inc: 773 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 774 // CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 775 // CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 776 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 777 // CHECK3: omp.inner.for.end: 778 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 779 // CHECK3: omp.loop.exit: 780 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 781 // CHECK3-NEXT: ret void 782 // 783 // 784 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33 785 // CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 786 // CHECK3-NEXT: entry: 787 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 788 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 789 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 790 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33.omp_outlined, ptr [[TMP0]]) 791 // CHECK3-NEXT: ret void 792 // 793 // 794 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l33.omp_outlined 795 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 796 // CHECK3-NEXT: entry: 797 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 798 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 799 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 800 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 801 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 802 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 803 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 804 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 805 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 806 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 807 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 808 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 809 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 810 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 811 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 812 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 813 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 814 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 815 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 816 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 817 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 818 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 819 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 820 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 821 // CHECK3: cond.true: 822 // CHECK3-NEXT: br label [[COND_END:%.*]] 823 // CHECK3: cond.false: 824 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 825 // CHECK3-NEXT: br label [[COND_END]] 826 // CHECK3: cond.end: 827 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 828 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 829 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 830 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 831 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 832 // CHECK3: omp.inner.for.cond: 833 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 834 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 835 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 836 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 837 // CHECK3: omp.inner.for.body: 838 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 839 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 840 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 841 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4 842 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 843 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4 844 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP9]] 845 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 846 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 847 // CHECK3: omp.body.continue: 848 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 849 // CHECK3: omp.inner.for.inc: 850 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 851 // CHECK3-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 852 // CHECK3-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 853 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 854 // CHECK3: omp.inner.for.end: 855 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 856 // CHECK3: omp.loop.exit: 857 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 858 // CHECK3-NEXT: ret void 859 // 860 // 861 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38 862 // CHECK3-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 863 // CHECK3-NEXT: entry: 864 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 865 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 866 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 867 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38.omp_outlined, ptr [[TMP0]]) 868 // CHECK3-NEXT: ret void 869 // 870 // 871 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l38.omp_outlined 872 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 873 // CHECK3-NEXT: entry: 874 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 875 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 876 // CHECK3-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 877 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 878 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 879 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 880 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 881 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 882 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 883 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 884 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 885 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 886 // CHECK3-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 887 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 888 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 889 // CHECK3-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 890 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 891 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 892 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 893 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 894 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 61) 895 // CHECK3-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 896 // CHECK3: omp.dispatch.cond: 897 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 898 // CHECK3-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 899 // CHECK3-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 900 // CHECK3: cond.true: 901 // CHECK3-NEXT: br label [[COND_END:%.*]] 902 // CHECK3: cond.false: 903 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 904 // CHECK3-NEXT: br label [[COND_END]] 905 // CHECK3: cond.end: 906 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 907 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 908 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 909 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 910 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 911 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 912 // CHECK3-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 913 // CHECK3-NEXT: br i1 [[CMP1]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 914 // CHECK3: omp.dispatch.body: 915 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 916 // CHECK3: omp.inner.for.cond: 917 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9:![0-9]+]] 918 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP9]] 919 // CHECK3-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 920 // CHECK3-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 921 // CHECK3: omp.inner.for.body: 922 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9]] 923 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 924 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 925 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP9]] 926 // CHECK3-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 927 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP9]] 928 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP11]] 929 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP9]] 930 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 931 // CHECK3: omp.body.continue: 932 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 933 // CHECK3: omp.inner.for.inc: 934 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9]] 935 // CHECK3-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1 936 // CHECK3-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9]] 937 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]] 938 // CHECK3: omp.inner.for.end: 939 // CHECK3-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 940 // CHECK3: omp.dispatch.inc: 941 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 942 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 943 // CHECK3-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP13]], [[TMP14]] 944 // CHECK3-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_LB]], align 4 945 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 946 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 947 // CHECK3-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP15]], [[TMP16]] 948 // CHECK3-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4 949 // CHECK3-NEXT: br label [[OMP_DISPATCH_COND]] 950 // CHECK3: omp.dispatch.end: 951 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 952 // CHECK3-NEXT: ret void 953 // 954 // 955 // CHECK9-LABEL: define {{[^@]+}}@main 956 // CHECK9-SAME: (i32 noundef signext [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 957 // CHECK9-NEXT: entry: 958 // CHECK9-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 959 // CHECK9-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 960 // CHECK9-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 8 961 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 962 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 8 963 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 964 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 965 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8 966 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8 967 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8 968 // CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8 969 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 970 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 971 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 972 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 973 // CHECK9-NEXT: [[N_CASTED3:%.*]] = alloca i64, align 8 974 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x ptr], align 8 975 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x ptr], align 8 976 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x ptr], align 8 977 // CHECK9-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 8 978 // CHECK9-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 979 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 980 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 981 // CHECK9-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 982 // CHECK9-NEXT: [[N_CASTED18:%.*]] = alloca i64, align 8 983 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [3 x ptr], align 8 984 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [3 x ptr], align 8 985 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [3 x ptr], align 8 986 // CHECK9-NEXT: [[DOTOFFLOAD_SIZES22:%.*]] = alloca [3 x i64], align 8 987 // CHECK9-NEXT: [[_TMP23:%.*]] = alloca i32, align 4 988 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4 989 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4 990 // CHECK9-NEXT: [[KERNEL_ARGS30:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 991 // CHECK9-NEXT: store i32 0, ptr [[RETVAL]], align 4 992 // CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 993 // CHECK9-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8 994 // CHECK9-NEXT: store i32 100, ptr [[N]], align 4 995 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 996 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 997 // CHECK9-NEXT: [[TMP2:%.*]] = call ptr @llvm.stacksave.p0() 998 // CHECK9-NEXT: store ptr [[TMP2]], ptr [[SAVED_STACK]], align 8 999 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 1000 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8 1001 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N]], align 4 1002 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 1003 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8 1004 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 1005 // CHECK9-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 24, i1 false) 1006 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1007 // CHECK9-NEXT: store i64 [[TMP4]], ptr [[TMP6]], align 8 1008 // CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1009 // CHECK9-NEXT: store i64 [[TMP4]], ptr [[TMP7]], align 8 1010 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 1011 // CHECK9-NEXT: store ptr null, ptr [[TMP8]], align 8 1012 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 1013 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP9]], align 8 1014 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 1015 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP10]], align 8 1016 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 1017 // CHECK9-NEXT: store ptr null, ptr [[TMP11]], align 8 1018 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 1019 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP12]], align 8 1020 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 1021 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP13]], align 8 1022 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2 1023 // CHECK9-NEXT: store i64 [[TMP5]], ptr [[TMP14]], align 8 1024 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 1025 // CHECK9-NEXT: store ptr null, ptr [[TMP15]], align 8 1026 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1027 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1028 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 1029 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, ptr [[N]], align 4 1030 // CHECK9-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4 1031 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1032 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0 1033 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1034 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1035 // CHECK9-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1036 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1037 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], 1 1038 // CHECK9-NEXT: [[TMP22:%.*]] = zext i32 [[ADD]] to i64 1039 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 1040 // CHECK9-NEXT: store i32 3, ptr [[TMP23]], align 4 1041 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 1042 // CHECK9-NEXT: store i32 3, ptr [[TMP24]], align 4 1043 // CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 1044 // CHECK9-NEXT: store ptr [[TMP16]], ptr [[TMP25]], align 8 1045 // CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 1046 // CHECK9-NEXT: store ptr [[TMP17]], ptr [[TMP26]], align 8 1047 // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 1048 // CHECK9-NEXT: store ptr [[TMP18]], ptr [[TMP27]], align 8 1049 // CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 1050 // CHECK9-NEXT: store ptr @.offload_maptypes, ptr [[TMP28]], align 8 1051 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 1052 // CHECK9-NEXT: store ptr null, ptr [[TMP29]], align 8 1053 // CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 1054 // CHECK9-NEXT: store ptr null, ptr [[TMP30]], align 8 1055 // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 1056 // CHECK9-NEXT: store i64 [[TMP22]], ptr [[TMP31]], align 8 1057 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 1058 // CHECK9-NEXT: store i64 0, ptr [[TMP32]], align 8 1059 // CHECK9-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 1060 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4 1061 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 1062 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP34]], align 4 1063 // CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 1064 // CHECK9-NEXT: store i32 0, ptr [[TMP35]], align 4 1065 // CHECK9-NEXT: [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, ptr [[KERNEL_ARGS]]) 1066 // CHECK9-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0 1067 // CHECK9-NEXT: br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1068 // CHECK9: omp_offload.failed: 1069 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i64 [[TMP4]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3:[0-9]+]] 1070 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] 1071 // CHECK9: omp_offload.cont: 1072 // CHECK9-NEXT: [[TMP38:%.*]] = load i32, ptr [[N]], align 4 1073 // CHECK9-NEXT: store i32 [[TMP38]], ptr [[N_CASTED3]], align 4 1074 // CHECK9-NEXT: [[TMP39:%.*]] = load i64, ptr [[N_CASTED3]], align 8 1075 // CHECK9-NEXT: [[TMP40:%.*]] = mul nuw i64 [[TMP1]], 4 1076 // CHECK9-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES7]], ptr align 8 @.offload_sizes.1, i64 24, i1 false) 1077 // CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 1078 // CHECK9-NEXT: store i64 [[TMP39]], ptr [[TMP41]], align 8 1079 // CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 1080 // CHECK9-NEXT: store i64 [[TMP39]], ptr [[TMP42]], align 8 1081 // CHECK9-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i64 0, i64 0 1082 // CHECK9-NEXT: store ptr null, ptr [[TMP43]], align 8 1083 // CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 1084 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP44]], align 8 1085 // CHECK9-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 1086 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP45]], align 8 1087 // CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i64 0, i64 1 1088 // CHECK9-NEXT: store ptr null, ptr [[TMP46]], align 8 1089 // CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 1090 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP47]], align 8 1091 // CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 1092 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP48]], align 8 1093 // CHECK9-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 1094 // CHECK9-NEXT: store i64 [[TMP40]], ptr [[TMP49]], align 8 1095 // CHECK9-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i64 0, i64 2 1096 // CHECK9-NEXT: store ptr null, ptr [[TMP50]], align 8 1097 // CHECK9-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 1098 // CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 1099 // CHECK9-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 1100 // CHECK9-NEXT: [[TMP54:%.*]] = load i32, ptr [[N]], align 4 1101 // CHECK9-NEXT: store i32 [[TMP54]], ptr [[DOTCAPTURE_EXPR_9]], align 4 1102 // CHECK9-NEXT: [[TMP55:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_9]], align 4 1103 // CHECK9-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP55]], 0 1104 // CHECK9-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 1105 // CHECK9-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 1106 // CHECK9-NEXT: store i32 [[SUB13]], ptr [[DOTCAPTURE_EXPR_10]], align 4 1107 // CHECK9-NEXT: [[TMP56:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_10]], align 4 1108 // CHECK9-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP56]], 1 1109 // CHECK9-NEXT: [[TMP57:%.*]] = zext i32 [[ADD14]] to i64 1110 // CHECK9-NEXT: [[TMP58:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0 1111 // CHECK9-NEXT: store i32 3, ptr [[TMP58]], align 4 1112 // CHECK9-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1 1113 // CHECK9-NEXT: store i32 3, ptr [[TMP59]], align 4 1114 // CHECK9-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2 1115 // CHECK9-NEXT: store ptr [[TMP51]], ptr [[TMP60]], align 8 1116 // CHECK9-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3 1117 // CHECK9-NEXT: store ptr [[TMP52]], ptr [[TMP61]], align 8 1118 // CHECK9-NEXT: [[TMP62:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4 1119 // CHECK9-NEXT: store ptr [[TMP53]], ptr [[TMP62]], align 8 1120 // CHECK9-NEXT: [[TMP63:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5 1121 // CHECK9-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP63]], align 8 1122 // CHECK9-NEXT: [[TMP64:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6 1123 // CHECK9-NEXT: store ptr null, ptr [[TMP64]], align 8 1124 // CHECK9-NEXT: [[TMP65:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7 1125 // CHECK9-NEXT: store ptr null, ptr [[TMP65]], align 8 1126 // CHECK9-NEXT: [[TMP66:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8 1127 // CHECK9-NEXT: store i64 [[TMP57]], ptr [[TMP66]], align 8 1128 // CHECK9-NEXT: [[TMP67:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9 1129 // CHECK9-NEXT: store i64 0, ptr [[TMP67]], align 8 1130 // CHECK9-NEXT: [[TMP68:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10 1131 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP68]], align 4 1132 // CHECK9-NEXT: [[TMP69:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11 1133 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP69]], align 4 1134 // CHECK9-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12 1135 // CHECK9-NEXT: store i32 0, ptr [[TMP70]], align 4 1136 // CHECK9-NEXT: [[TMP71:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, ptr [[KERNEL_ARGS15]]) 1137 // CHECK9-NEXT: [[TMP72:%.*]] = icmp ne i32 [[TMP71]], 0 1138 // CHECK9-NEXT: br i1 [[TMP72]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] 1139 // CHECK9: omp_offload.failed16: 1140 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i64 [[TMP39]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3]] 1141 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT17]] 1142 // CHECK9: omp_offload.cont17: 1143 // CHECK9-NEXT: [[TMP73:%.*]] = load i32, ptr [[N]], align 4 1144 // CHECK9-NEXT: store i32 [[TMP73]], ptr [[N_CASTED18]], align 4 1145 // CHECK9-NEXT: [[TMP74:%.*]] = load i64, ptr [[N_CASTED18]], align 8 1146 // CHECK9-NEXT: [[TMP75:%.*]] = mul nuw i64 [[TMP1]], 4 1147 // CHECK9-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES22]], ptr align 8 @.offload_sizes.3, i64 24, i1 false) 1148 // CHECK9-NEXT: [[TMP76:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 1149 // CHECK9-NEXT: store i64 [[TMP74]], ptr [[TMP76]], align 8 1150 // CHECK9-NEXT: [[TMP77:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 1151 // CHECK9-NEXT: store i64 [[TMP74]], ptr [[TMP77]], align 8 1152 // CHECK9-NEXT: [[TMP78:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 0 1153 // CHECK9-NEXT: store ptr null, ptr [[TMP78]], align 8 1154 // CHECK9-NEXT: [[TMP79:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1 1155 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP79]], align 8 1156 // CHECK9-NEXT: [[TMP80:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 1 1157 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP80]], align 8 1158 // CHECK9-NEXT: [[TMP81:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 1 1159 // CHECK9-NEXT: store ptr null, ptr [[TMP81]], align 8 1160 // CHECK9-NEXT: [[TMP82:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2 1161 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP82]], align 8 1162 // CHECK9-NEXT: [[TMP83:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 2 1163 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP83]], align 8 1164 // CHECK9-NEXT: [[TMP84:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES22]], i32 0, i32 2 1165 // CHECK9-NEXT: store i64 [[TMP75]], ptr [[TMP84]], align 8 1166 // CHECK9-NEXT: [[TMP85:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i64 0, i64 2 1167 // CHECK9-NEXT: store ptr null, ptr [[TMP85]], align 8 1168 // CHECK9-NEXT: [[TMP86:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 1169 // CHECK9-NEXT: [[TMP87:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 1170 // CHECK9-NEXT: [[TMP88:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES22]], i32 0, i32 0 1171 // CHECK9-NEXT: [[TMP89:%.*]] = load i32, ptr [[N]], align 4 1172 // CHECK9-NEXT: store i32 [[TMP89]], ptr [[DOTCAPTURE_EXPR_24]], align 4 1173 // CHECK9-NEXT: [[TMP90:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_24]], align 4 1174 // CHECK9-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP90]], 0 1175 // CHECK9-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1 1176 // CHECK9-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1 1177 // CHECK9-NEXT: store i32 [[SUB28]], ptr [[DOTCAPTURE_EXPR_25]], align 4 1178 // CHECK9-NEXT: [[TMP91:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_25]], align 4 1179 // CHECK9-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP91]], 1 1180 // CHECK9-NEXT: [[TMP92:%.*]] = zext i32 [[ADD29]] to i64 1181 // CHECK9-NEXT: [[TMP93:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 0 1182 // CHECK9-NEXT: store i32 3, ptr [[TMP93]], align 4 1183 // CHECK9-NEXT: [[TMP94:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 1 1184 // CHECK9-NEXT: store i32 3, ptr [[TMP94]], align 4 1185 // CHECK9-NEXT: [[TMP95:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 2 1186 // CHECK9-NEXT: store ptr [[TMP86]], ptr [[TMP95]], align 8 1187 // CHECK9-NEXT: [[TMP96:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 3 1188 // CHECK9-NEXT: store ptr [[TMP87]], ptr [[TMP96]], align 8 1189 // CHECK9-NEXT: [[TMP97:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 4 1190 // CHECK9-NEXT: store ptr [[TMP88]], ptr [[TMP97]], align 8 1191 // CHECK9-NEXT: [[TMP98:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 5 1192 // CHECK9-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP98]], align 8 1193 // CHECK9-NEXT: [[TMP99:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 6 1194 // CHECK9-NEXT: store ptr null, ptr [[TMP99]], align 8 1195 // CHECK9-NEXT: [[TMP100:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 7 1196 // CHECK9-NEXT: store ptr null, ptr [[TMP100]], align 8 1197 // CHECK9-NEXT: [[TMP101:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 8 1198 // CHECK9-NEXT: store i64 [[TMP92]], ptr [[TMP101]], align 8 1199 // CHECK9-NEXT: [[TMP102:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 9 1200 // CHECK9-NEXT: store i64 0, ptr [[TMP102]], align 8 1201 // CHECK9-NEXT: [[TMP103:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 10 1202 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP103]], align 4 1203 // CHECK9-NEXT: [[TMP104:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 11 1204 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP104]], align 4 1205 // CHECK9-NEXT: [[TMP105:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 12 1206 // CHECK9-NEXT: store i32 0, ptr [[TMP105]], align 4 1207 // CHECK9-NEXT: [[TMP106:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, ptr [[KERNEL_ARGS30]]) 1208 // CHECK9-NEXT: [[TMP107:%.*]] = icmp ne i32 [[TMP106]], 0 1209 // CHECK9-NEXT: br i1 [[TMP107]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]] 1210 // CHECK9: omp_offload.failed31: 1211 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i64 [[TMP74]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3]] 1212 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT32]] 1213 // CHECK9: omp_offload.cont32: 1214 // CHECK9-NEXT: [[TMP108:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 1215 // CHECK9-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10EEiT_(i32 noundef signext [[TMP108]]) 1216 // CHECK9-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 1217 // CHECK9-NEXT: [[TMP109:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 1218 // CHECK9-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP109]]) 1219 // CHECK9-NEXT: [[TMP110:%.*]] = load i32, ptr [[RETVAL]], align 4 1220 // CHECK9-NEXT: ret i32 [[TMP110]] 1221 // 1222 // 1223 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 1224 // CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 1225 // CHECK9-NEXT: entry: 1226 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 1227 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1228 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1229 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 1230 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 1231 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1232 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 1233 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1234 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]]) 1235 // CHECK9-NEXT: ret void 1236 // 1237 // 1238 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.omp_outlined 1239 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1240 // CHECK9-NEXT: entry: 1241 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 1242 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 1243 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8 1244 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1245 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1246 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1247 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1248 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1249 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1250 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 1251 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1252 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1253 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1254 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1255 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 1256 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 1257 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 1258 // CHECK9-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8 1259 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 1260 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1261 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8 1262 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 1263 // CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1264 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 1265 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 1266 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1267 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1268 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1269 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1270 // CHECK9-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1271 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4 1272 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1273 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1274 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1275 // CHECK9: omp.precond.then: 1276 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1277 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1278 // CHECK9-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 1279 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1280 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1281 // CHECK9-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1282 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 1283 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1284 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1285 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1286 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 1287 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1288 // CHECK9: cond.true: 1289 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1290 // CHECK9-NEXT: br label [[COND_END:%.*]] 1291 // CHECK9: cond.false: 1292 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1293 // CHECK9-NEXT: br label [[COND_END]] 1294 // CHECK9: cond.end: 1295 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 1296 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 1297 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1298 // CHECK9-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 1299 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1300 // CHECK9: omp.inner.for.cond: 1301 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1302 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1303 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 1304 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1305 // CHECK9: omp.inner.for.body: 1306 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1307 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 1308 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1309 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I3]], align 4 1310 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, ptr [[I3]], align 4 1311 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP17]] to i64 1312 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]] 1313 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 1314 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1315 // CHECK9: omp.body.continue: 1316 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1317 // CHECK9: omp.inner.for.inc: 1318 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1319 // CHECK9-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], 1 1320 // CHECK9-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 1321 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 1322 // CHECK9: omp.inner.for.end: 1323 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1324 // CHECK9: omp.loop.exit: 1325 // CHECK9-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1326 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4 1327 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP20]]) 1328 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 1329 // CHECK9: omp.precond.end: 1330 // CHECK9-NEXT: ret void 1331 // 1332 // 1333 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105 1334 // CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1335 // CHECK9-NEXT: entry: 1336 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 1337 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1338 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1339 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 1340 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 1341 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1342 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 1343 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1344 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]]) 1345 // CHECK9-NEXT: ret void 1346 // 1347 // 1348 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.omp_outlined 1349 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1350 // CHECK9-NEXT: entry: 1351 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 1352 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 1353 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8 1354 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1355 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1356 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1357 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1358 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1359 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1360 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 1361 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1362 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1363 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1364 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1365 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 1366 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 1367 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 1368 // CHECK9-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8 1369 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 1370 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1371 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8 1372 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 1373 // CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1374 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 1375 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 1376 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1377 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1378 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1379 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1380 // CHECK9-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1381 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4 1382 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1383 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1384 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1385 // CHECK9: omp.precond.then: 1386 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1387 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1388 // CHECK9-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 1389 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1390 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1391 // CHECK9-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1392 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 1393 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1394 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1395 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1396 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 1397 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1398 // CHECK9: cond.true: 1399 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1400 // CHECK9-NEXT: br label [[COND_END:%.*]] 1401 // CHECK9: cond.false: 1402 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1403 // CHECK9-NEXT: br label [[COND_END]] 1404 // CHECK9: cond.end: 1405 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 1406 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 1407 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1408 // CHECK9-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 1409 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1410 // CHECK9: omp.inner.for.cond: 1411 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1412 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1413 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 1414 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1415 // CHECK9: omp.inner.for.body: 1416 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1417 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 1418 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1419 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I3]], align 4 1420 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, ptr [[I3]], align 4 1421 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP17]] to i64 1422 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]] 1423 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 1424 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1425 // CHECK9: omp.body.continue: 1426 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1427 // CHECK9: omp.inner.for.inc: 1428 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1429 // CHECK9-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], 1 1430 // CHECK9-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 1431 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 1432 // CHECK9: omp.inner.for.end: 1433 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1434 // CHECK9: omp.loop.exit: 1435 // CHECK9-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1436 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4 1437 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP20]]) 1438 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 1439 // CHECK9: omp.precond.end: 1440 // CHECK9-NEXT: ret void 1441 // 1442 // 1443 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110 1444 // CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1445 // CHECK9-NEXT: entry: 1446 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 1447 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1448 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1449 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1450 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 1451 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 1452 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 1453 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1454 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 1455 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1456 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 1457 // CHECK9-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 1458 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1459 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 1460 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8 1461 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]], i64 [[TMP4]]) 1462 // CHECK9-NEXT: ret void 1463 // 1464 // 1465 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.omp_outlined 1466 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 1467 // CHECK9-NEXT: entry: 1468 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 1469 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 1470 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8 1471 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1472 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1473 // CHECK9-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 1474 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1475 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1476 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1477 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 1478 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 1479 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1480 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1481 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1482 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1483 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 1484 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 1485 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 1486 // CHECK9-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8 1487 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 1488 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1489 // CHECK9-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8 1490 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8 1491 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 1492 // CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1493 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 1494 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1495 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1496 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1497 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1498 // CHECK9-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 1499 // CHECK9-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4 1500 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4 1501 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1502 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1503 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1504 // CHECK9: omp.precond.then: 1505 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1506 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 1507 // CHECK9-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 1508 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1509 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1510 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 1511 // CHECK9-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1512 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4 1513 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP7]]) 1514 // CHECK9-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 1515 // CHECK9: omp.dispatch.cond: 1516 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1517 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 1518 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 1519 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1520 // CHECK9: cond.true: 1521 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 1522 // CHECK9-NEXT: br label [[COND_END:%.*]] 1523 // CHECK9: cond.false: 1524 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1525 // CHECK9-NEXT: br label [[COND_END]] 1526 // CHECK9: cond.end: 1527 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 1528 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 1529 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1530 // CHECK9-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4 1531 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1532 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1533 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 1534 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 1535 // CHECK9: omp.dispatch.body: 1536 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1537 // CHECK9: omp.inner.for.cond: 1538 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11:![0-9]+]] 1539 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP11]] 1540 // CHECK9-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 1541 // CHECK9-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1542 // CHECK9: omp.inner.for.body: 1543 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]] 1544 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 1545 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1546 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP11]] 1547 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP11]] 1548 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP20]] to i64 1549 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]] 1550 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP11]] 1551 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1552 // CHECK9: omp.body.continue: 1553 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1554 // CHECK9: omp.inner.for.inc: 1555 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]] 1556 // CHECK9-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP21]], 1 1557 // CHECK9-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]] 1558 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]] 1559 // CHECK9: omp.inner.for.end: 1560 // CHECK9-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 1561 // CHECK9: omp.dispatch.inc: 1562 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1563 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 1564 // CHECK9-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP22]], [[TMP23]] 1565 // CHECK9-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_LB]], align 4 1566 // CHECK9-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1567 // CHECK9-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 1568 // CHECK9-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP24]], [[TMP25]] 1569 // CHECK9-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_UB]], align 4 1570 // CHECK9-NEXT: br label [[OMP_DISPATCH_COND]] 1571 // CHECK9: omp.dispatch.end: 1572 // CHECK9-NEXT: [[TMP26:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1573 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, ptr [[TMP26]], align 4 1574 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP27]]) 1575 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 1576 // CHECK9: omp.precond.end: 1577 // CHECK9-NEXT: ret void 1578 // 1579 // 1580 // CHECK9-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 1581 // CHECK9-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 1582 // CHECK9-NEXT: entry: 1583 // CHECK9-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 1584 // CHECK9-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 1585 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 1586 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 1587 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 1588 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1589 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 1590 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [1 x ptr], align 8 1591 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [1 x ptr], align 8 1592 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [1 x ptr], align 8 1593 // CHECK9-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 1594 // CHECK9-NEXT: [[KERNEL_ARGS5:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 1595 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS8:%.*]] = alloca [1 x ptr], align 8 1596 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS9:%.*]] = alloca [1 x ptr], align 8 1597 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS10:%.*]] = alloca [1 x ptr], align 8 1598 // CHECK9-NEXT: [[_TMP11:%.*]] = alloca i32, align 4 1599 // CHECK9-NEXT: [[KERNEL_ARGS12:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 1600 // CHECK9-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 1601 // CHECK9-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1602 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP0]], align 8 1603 // CHECK9-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1604 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP1]], align 8 1605 // CHECK9-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 1606 // CHECK9-NEXT: store ptr null, ptr [[TMP2]], align 8 1607 // CHECK9-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1608 // CHECK9-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1609 // CHECK9-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 1610 // CHECK9-NEXT: store i32 3, ptr [[TMP5]], align 4 1611 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 1612 // CHECK9-NEXT: store i32 1, ptr [[TMP6]], align 4 1613 // CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 1614 // CHECK9-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8 1615 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 1616 // CHECK9-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8 1617 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 1618 // CHECK9-NEXT: store ptr @.offload_sizes.5, ptr [[TMP9]], align 8 1619 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 1620 // CHECK9-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP10]], align 8 1621 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 1622 // CHECK9-NEXT: store ptr null, ptr [[TMP11]], align 8 1623 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 1624 // CHECK9-NEXT: store ptr null, ptr [[TMP12]], align 8 1625 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 1626 // CHECK9-NEXT: store i64 10, ptr [[TMP13]], align 8 1627 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 1628 // CHECK9-NEXT: store i64 0, ptr [[TMP14]], align 8 1629 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 1630 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4 1631 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 1632 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4 1633 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 1634 // CHECK9-NEXT: store i32 0, ptr [[TMP17]], align 4 1635 // CHECK9-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, ptr [[KERNEL_ARGS]]) 1636 // CHECK9-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0 1637 // CHECK9-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1638 // CHECK9: omp_offload.failed: 1639 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79(ptr [[A]]) #[[ATTR3]] 1640 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] 1641 // CHECK9: omp_offload.cont: 1642 // CHECK9-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 1643 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP20]], align 8 1644 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 1645 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP21]], align 8 1646 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i64 0, i64 0 1647 // CHECK9-NEXT: store ptr null, ptr [[TMP22]], align 8 1648 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 1649 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 1650 // CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 0 1651 // CHECK9-NEXT: store i32 3, ptr [[TMP25]], align 4 1652 // CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 1 1653 // CHECK9-NEXT: store i32 1, ptr [[TMP26]], align 4 1654 // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 2 1655 // CHECK9-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 8 1656 // CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 3 1657 // CHECK9-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 8 1658 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 4 1659 // CHECK9-NEXT: store ptr @.offload_sizes.7, ptr [[TMP29]], align 8 1660 // CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 5 1661 // CHECK9-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP30]], align 8 1662 // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 6 1663 // CHECK9-NEXT: store ptr null, ptr [[TMP31]], align 8 1664 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 7 1665 // CHECK9-NEXT: store ptr null, ptr [[TMP32]], align 8 1666 // CHECK9-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 8 1667 // CHECK9-NEXT: store i64 10, ptr [[TMP33]], align 8 1668 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 9 1669 // CHECK9-NEXT: store i64 0, ptr [[TMP34]], align 8 1670 // CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 10 1671 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4 1672 // CHECK9-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 11 1673 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4 1674 // CHECK9-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 12 1675 // CHECK9-NEXT: store i32 0, ptr [[TMP37]], align 4 1676 // CHECK9-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, ptr [[KERNEL_ARGS5]]) 1677 // CHECK9-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0 1678 // CHECK9-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]] 1679 // CHECK9: omp_offload.failed6: 1680 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84(ptr [[A]]) #[[ATTR3]] 1681 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT7]] 1682 // CHECK9: omp_offload.cont7: 1683 // CHECK9-NEXT: [[TMP40:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0 1684 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP40]], align 8 1685 // CHECK9-NEXT: [[TMP41:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0 1686 // CHECK9-NEXT: store ptr [[A]], ptr [[TMP41]], align 8 1687 // CHECK9-NEXT: [[TMP42:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i64 0, i64 0 1688 // CHECK9-NEXT: store ptr null, ptr [[TMP42]], align 8 1689 // CHECK9-NEXT: [[TMP43:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0 1690 // CHECK9-NEXT: [[TMP44:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0 1691 // CHECK9-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 0 1692 // CHECK9-NEXT: store i32 3, ptr [[TMP45]], align 4 1693 // CHECK9-NEXT: [[TMP46:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 1 1694 // CHECK9-NEXT: store i32 1, ptr [[TMP46]], align 4 1695 // CHECK9-NEXT: [[TMP47:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 2 1696 // CHECK9-NEXT: store ptr [[TMP43]], ptr [[TMP47]], align 8 1697 // CHECK9-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 3 1698 // CHECK9-NEXT: store ptr [[TMP44]], ptr [[TMP48]], align 8 1699 // CHECK9-NEXT: [[TMP49:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 4 1700 // CHECK9-NEXT: store ptr @.offload_sizes.9, ptr [[TMP49]], align 8 1701 // CHECK9-NEXT: [[TMP50:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 5 1702 // CHECK9-NEXT: store ptr @.offload_maptypes.10, ptr [[TMP50]], align 8 1703 // CHECK9-NEXT: [[TMP51:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 6 1704 // CHECK9-NEXT: store ptr null, ptr [[TMP51]], align 8 1705 // CHECK9-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 7 1706 // CHECK9-NEXT: store ptr null, ptr [[TMP52]], align 8 1707 // CHECK9-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 8 1708 // CHECK9-NEXT: store i64 10, ptr [[TMP53]], align 8 1709 // CHECK9-NEXT: [[TMP54:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 9 1710 // CHECK9-NEXT: store i64 0, ptr [[TMP54]], align 8 1711 // CHECK9-NEXT: [[TMP55:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 10 1712 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP55]], align 4 1713 // CHECK9-NEXT: [[TMP56:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 11 1714 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP56]], align 4 1715 // CHECK9-NEXT: [[TMP57:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 12 1716 // CHECK9-NEXT: store i32 0, ptr [[TMP57]], align 4 1717 // CHECK9-NEXT: [[TMP58:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, ptr [[KERNEL_ARGS12]]) 1718 // CHECK9-NEXT: [[TMP59:%.*]] = icmp ne i32 [[TMP58]], 0 1719 // CHECK9-NEXT: br i1 [[TMP59]], label [[OMP_OFFLOAD_FAILED13:%.*]], label [[OMP_OFFLOAD_CONT14:%.*]] 1720 // CHECK9: omp_offload.failed13: 1721 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89(ptr [[A]]) #[[ATTR3]] 1722 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT14]] 1723 // CHECK9: omp_offload.cont14: 1724 // CHECK9-NEXT: ret i32 0 1725 // 1726 // 1727 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79 1728 // CHECK9-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 1729 // CHECK9-NEXT: entry: 1730 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1731 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1732 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1733 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.omp_outlined, ptr [[TMP0]]) 1734 // CHECK9-NEXT: ret void 1735 // 1736 // 1737 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.omp_outlined 1738 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 1739 // CHECK9-NEXT: entry: 1740 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 1741 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 1742 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1743 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1744 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1745 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1746 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1747 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1748 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1749 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 1750 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 1751 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 1752 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1753 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1754 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1755 // CHECK9-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 1756 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1757 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1758 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1759 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 1760 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1761 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1762 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 1763 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1764 // CHECK9: cond.true: 1765 // CHECK9-NEXT: br label [[COND_END:%.*]] 1766 // CHECK9: cond.false: 1767 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1768 // CHECK9-NEXT: br label [[COND_END]] 1769 // CHECK9: cond.end: 1770 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 1771 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 1772 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1773 // CHECK9-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 1774 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1775 // CHECK9: omp.inner.for.cond: 1776 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1777 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1778 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 1779 // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1780 // CHECK9: omp.inner.for.body: 1781 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1782 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 1783 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1784 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4 1785 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4 1786 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP9]] to i64 1787 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]] 1788 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 1789 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1790 // CHECK9: omp.body.continue: 1791 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1792 // CHECK9: omp.inner.for.inc: 1793 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1794 // CHECK9-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 1795 // CHECK9-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 1796 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 1797 // CHECK9: omp.inner.for.end: 1798 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1799 // CHECK9: omp.loop.exit: 1800 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 1801 // CHECK9-NEXT: ret void 1802 // 1803 // 1804 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 1805 // CHECK9-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 1806 // CHECK9-NEXT: entry: 1807 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1808 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1809 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1810 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.omp_outlined, ptr [[TMP0]]) 1811 // CHECK9-NEXT: ret void 1812 // 1813 // 1814 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.omp_outlined 1815 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 1816 // CHECK9-NEXT: entry: 1817 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 1818 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 1819 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1820 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1821 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1822 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1823 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1824 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1825 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1826 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 1827 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 1828 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 1829 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1830 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1831 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1832 // CHECK9-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 1833 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1834 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1835 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1836 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 1837 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1838 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1839 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 1840 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1841 // CHECK9: cond.true: 1842 // CHECK9-NEXT: br label [[COND_END:%.*]] 1843 // CHECK9: cond.false: 1844 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1845 // CHECK9-NEXT: br label [[COND_END]] 1846 // CHECK9: cond.end: 1847 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 1848 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 1849 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1850 // CHECK9-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 1851 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1852 // CHECK9: omp.inner.for.cond: 1853 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1854 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1855 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 1856 // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1857 // CHECK9: omp.inner.for.body: 1858 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1859 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 1860 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1861 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4 1862 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4 1863 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP9]] to i64 1864 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]] 1865 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 1866 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1867 // CHECK9: omp.body.continue: 1868 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1869 // CHECK9: omp.inner.for.inc: 1870 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1871 // CHECK9-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 1872 // CHECK9-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 1873 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 1874 // CHECK9: omp.inner.for.end: 1875 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1876 // CHECK9: omp.loop.exit: 1877 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 1878 // CHECK9-NEXT: ret void 1879 // 1880 // 1881 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89 1882 // CHECK9-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 1883 // CHECK9-NEXT: entry: 1884 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1885 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1886 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1887 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined, ptr [[TMP0]]) 1888 // CHECK9-NEXT: ret void 1889 // 1890 // 1891 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined 1892 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 1893 // CHECK9-NEXT: entry: 1894 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 1895 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 1896 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1897 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1898 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1899 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1900 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1901 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1902 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1903 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 1904 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 1905 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 1906 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1907 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1908 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1909 // CHECK9-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 1910 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1911 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1912 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1913 // CHECK9-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 1914 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 10) 1915 // CHECK9-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 1916 // CHECK9: omp.dispatch.cond: 1917 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1918 // CHECK9-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 1919 // CHECK9-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1920 // CHECK9: cond.true: 1921 // CHECK9-NEXT: br label [[COND_END:%.*]] 1922 // CHECK9: cond.false: 1923 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1924 // CHECK9-NEXT: br label [[COND_END]] 1925 // CHECK9: cond.end: 1926 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 1927 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 1928 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1929 // CHECK9-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 1930 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1931 // CHECK9-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1932 // CHECK9-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 1933 // CHECK9-NEXT: br i1 [[CMP1]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 1934 // CHECK9: omp.dispatch.body: 1935 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1936 // CHECK9: omp.inner.for.cond: 1937 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14:![0-9]+]] 1938 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP14]] 1939 // CHECK9-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 1940 // CHECK9-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1941 // CHECK9: omp.inner.for.body: 1942 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14]] 1943 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 1944 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1945 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP14]] 1946 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP14]] 1947 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 1948 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]] 1949 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP14]] 1950 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1951 // CHECK9: omp.body.continue: 1952 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1953 // CHECK9: omp.inner.for.inc: 1954 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14]] 1955 // CHECK9-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1 1956 // CHECK9-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14]] 1957 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]] 1958 // CHECK9: omp.inner.for.end: 1959 // CHECK9-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 1960 // CHECK9: omp.dispatch.inc: 1961 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1962 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 1963 // CHECK9-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP13]], [[TMP14]] 1964 // CHECK9-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_LB]], align 4 1965 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1966 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 1967 // CHECK9-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP15]], [[TMP16]] 1968 // CHECK9-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4 1969 // CHECK9-NEXT: br label [[OMP_DISPATCH_COND]] 1970 // CHECK9: omp.dispatch.end: 1971 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 1972 // CHECK9-NEXT: ret void 1973 // 1974 // 1975 // CHECK11-LABEL: define {{[^@]+}}@main 1976 // CHECK11-SAME: (i32 noundef [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 1977 // CHECK11-NEXT: entry: 1978 // CHECK11-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 1979 // CHECK11-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 1980 // CHECK11-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 4 1981 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 1982 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 4 1983 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 1984 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 1985 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4 1986 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4 1987 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4 1988 // CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4 1989 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 1990 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1991 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1992 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 1993 // CHECK11-NEXT: [[N_CASTED3:%.*]] = alloca i32, align 4 1994 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS4:%.*]] = alloca [3 x ptr], align 4 1995 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS5:%.*]] = alloca [3 x ptr], align 4 1996 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS6:%.*]] = alloca [3 x ptr], align 4 1997 // CHECK11-NEXT: [[DOTOFFLOAD_SIZES7:%.*]] = alloca [3 x i64], align 4 1998 // CHECK11-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 1999 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 2000 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 2001 // CHECK11-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 2002 // CHECK11-NEXT: [[N_CASTED18:%.*]] = alloca i32, align 4 2003 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS19:%.*]] = alloca [3 x ptr], align 4 2004 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS20:%.*]] = alloca [3 x ptr], align 4 2005 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS21:%.*]] = alloca [3 x ptr], align 4 2006 // CHECK11-NEXT: [[DOTOFFLOAD_SIZES22:%.*]] = alloca [3 x i64], align 4 2007 // CHECK11-NEXT: [[_TMP23:%.*]] = alloca i32, align 4 2008 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_24:%.*]] = alloca i32, align 4 2009 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_25:%.*]] = alloca i32, align 4 2010 // CHECK11-NEXT: [[KERNEL_ARGS30:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 2011 // CHECK11-NEXT: store i32 0, ptr [[RETVAL]], align 4 2012 // CHECK11-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 2013 // CHECK11-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 4 2014 // CHECK11-NEXT: store i32 100, ptr [[N]], align 4 2015 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 2016 // CHECK11-NEXT: [[TMP1:%.*]] = call ptr @llvm.stacksave.p0() 2017 // CHECK11-NEXT: store ptr [[TMP1]], ptr [[SAVED_STACK]], align 4 2018 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 2019 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[__VLA_EXPR0]], align 4 2020 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[N]], align 4 2021 // CHECK11-NEXT: store i32 [[TMP2]], ptr [[N_CASTED]], align 4 2022 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_CASTED]], align 4 2023 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 2024 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 2025 // CHECK11-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 24, i1 false) 2026 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2027 // CHECK11-NEXT: store i32 [[TMP3]], ptr [[TMP6]], align 4 2028 // CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2029 // CHECK11-NEXT: store i32 [[TMP3]], ptr [[TMP7]], align 4 2030 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 2031 // CHECK11-NEXT: store ptr null, ptr [[TMP8]], align 4 2032 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 2033 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP9]], align 4 2034 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 2035 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP10]], align 4 2036 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 2037 // CHECK11-NEXT: store ptr null, ptr [[TMP11]], align 4 2038 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 2039 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP12]], align 4 2040 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 2041 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP13]], align 4 2042 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2 2043 // CHECK11-NEXT: store i64 [[TMP5]], ptr [[TMP14]], align 4 2044 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 2045 // CHECK11-NEXT: store ptr null, ptr [[TMP15]], align 4 2046 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2047 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2048 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 2049 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, ptr [[N]], align 4 2050 // CHECK11-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4 2051 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 2052 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0 2053 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2054 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 2055 // CHECK11-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 2056 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2057 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], 1 2058 // CHECK11-NEXT: [[TMP22:%.*]] = zext i32 [[ADD]] to i64 2059 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 2060 // CHECK11-NEXT: store i32 3, ptr [[TMP23]], align 4 2061 // CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 2062 // CHECK11-NEXT: store i32 3, ptr [[TMP24]], align 4 2063 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 2064 // CHECK11-NEXT: store ptr [[TMP16]], ptr [[TMP25]], align 4 2065 // CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 2066 // CHECK11-NEXT: store ptr [[TMP17]], ptr [[TMP26]], align 4 2067 // CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 2068 // CHECK11-NEXT: store ptr [[TMP18]], ptr [[TMP27]], align 4 2069 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 2070 // CHECK11-NEXT: store ptr @.offload_maptypes, ptr [[TMP28]], align 4 2071 // CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 2072 // CHECK11-NEXT: store ptr null, ptr [[TMP29]], align 4 2073 // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 2074 // CHECK11-NEXT: store ptr null, ptr [[TMP30]], align 4 2075 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 2076 // CHECK11-NEXT: store i64 [[TMP22]], ptr [[TMP31]], align 8 2077 // CHECK11-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 2078 // CHECK11-NEXT: store i64 0, ptr [[TMP32]], align 8 2079 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 2080 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4 2081 // CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 2082 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP34]], align 4 2083 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 2084 // CHECK11-NEXT: store i32 0, ptr [[TMP35]], align 4 2085 // CHECK11-NEXT: [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.region_id, ptr [[KERNEL_ARGS]]) 2086 // CHECK11-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0 2087 // CHECK11-NEXT: br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2088 // CHECK11: omp_offload.failed: 2089 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100(i32 [[TMP3]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3:[0-9]+]] 2090 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] 2091 // CHECK11: omp_offload.cont: 2092 // CHECK11-NEXT: [[TMP38:%.*]] = load i32, ptr [[N]], align 4 2093 // CHECK11-NEXT: store i32 [[TMP38]], ptr [[N_CASTED3]], align 4 2094 // CHECK11-NEXT: [[TMP39:%.*]] = load i32, ptr [[N_CASTED3]], align 4 2095 // CHECK11-NEXT: [[TMP40:%.*]] = mul nuw i32 [[TMP0]], 4 2096 // CHECK11-NEXT: [[TMP41:%.*]] = sext i32 [[TMP40]] to i64 2097 // CHECK11-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES7]], ptr align 4 @.offload_sizes.1, i32 24, i1 false) 2098 // CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 2099 // CHECK11-NEXT: store i32 [[TMP39]], ptr [[TMP42]], align 4 2100 // CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 2101 // CHECK11-NEXT: store i32 [[TMP39]], ptr [[TMP43]], align 4 2102 // CHECK11-NEXT: [[TMP44:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 0 2103 // CHECK11-NEXT: store ptr null, ptr [[TMP44]], align 4 2104 // CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 1 2105 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP45]], align 4 2106 // CHECK11-NEXT: [[TMP46:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 1 2107 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP46]], align 4 2108 // CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 1 2109 // CHECK11-NEXT: store ptr null, ptr [[TMP47]], align 4 2110 // CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 2 2111 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP48]], align 4 2112 // CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 2 2113 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP49]], align 4 2114 // CHECK11-NEXT: [[TMP50:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 2 2115 // CHECK11-NEXT: store i64 [[TMP41]], ptr [[TMP50]], align 4 2116 // CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS6]], i32 0, i32 2 2117 // CHECK11-NEXT: store ptr null, ptr [[TMP51]], align 4 2118 // CHECK11-NEXT: [[TMP52:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS4]], i32 0, i32 0 2119 // CHECK11-NEXT: [[TMP53:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS5]], i32 0, i32 0 2120 // CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES7]], i32 0, i32 0 2121 // CHECK11-NEXT: [[TMP55:%.*]] = load i32, ptr [[N]], align 4 2122 // CHECK11-NEXT: store i32 [[TMP55]], ptr [[DOTCAPTURE_EXPR_9]], align 4 2123 // CHECK11-NEXT: [[TMP56:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_9]], align 4 2124 // CHECK11-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP56]], 0 2125 // CHECK11-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 2126 // CHECK11-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 2127 // CHECK11-NEXT: store i32 [[SUB13]], ptr [[DOTCAPTURE_EXPR_10]], align 4 2128 // CHECK11-NEXT: [[TMP57:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_10]], align 4 2129 // CHECK11-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP57]], 1 2130 // CHECK11-NEXT: [[TMP58:%.*]] = zext i32 [[ADD14]] to i64 2131 // CHECK11-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0 2132 // CHECK11-NEXT: store i32 3, ptr [[TMP59]], align 4 2133 // CHECK11-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1 2134 // CHECK11-NEXT: store i32 3, ptr [[TMP60]], align 4 2135 // CHECK11-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2 2136 // CHECK11-NEXT: store ptr [[TMP52]], ptr [[TMP61]], align 4 2137 // CHECK11-NEXT: [[TMP62:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3 2138 // CHECK11-NEXT: store ptr [[TMP53]], ptr [[TMP62]], align 4 2139 // CHECK11-NEXT: [[TMP63:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4 2140 // CHECK11-NEXT: store ptr [[TMP54]], ptr [[TMP63]], align 4 2141 // CHECK11-NEXT: [[TMP64:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5 2142 // CHECK11-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP64]], align 4 2143 // CHECK11-NEXT: [[TMP65:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6 2144 // CHECK11-NEXT: store ptr null, ptr [[TMP65]], align 4 2145 // CHECK11-NEXT: [[TMP66:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7 2146 // CHECK11-NEXT: store ptr null, ptr [[TMP66]], align 4 2147 // CHECK11-NEXT: [[TMP67:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8 2148 // CHECK11-NEXT: store i64 [[TMP58]], ptr [[TMP67]], align 8 2149 // CHECK11-NEXT: [[TMP68:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9 2150 // CHECK11-NEXT: store i64 0, ptr [[TMP68]], align 8 2151 // CHECK11-NEXT: [[TMP69:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10 2152 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP69]], align 4 2153 // CHECK11-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11 2154 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP70]], align 4 2155 // CHECK11-NEXT: [[TMP71:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12 2156 // CHECK11-NEXT: store i32 0, ptr [[TMP71]], align 4 2157 // CHECK11-NEXT: [[TMP72:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.region_id, ptr [[KERNEL_ARGS15]]) 2158 // CHECK11-NEXT: [[TMP73:%.*]] = icmp ne i32 [[TMP72]], 0 2159 // CHECK11-NEXT: br i1 [[TMP73]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] 2160 // CHECK11: omp_offload.failed16: 2161 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105(i32 [[TMP39]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3]] 2162 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT17]] 2163 // CHECK11: omp_offload.cont17: 2164 // CHECK11-NEXT: [[TMP74:%.*]] = load i32, ptr [[N]], align 4 2165 // CHECK11-NEXT: store i32 [[TMP74]], ptr [[N_CASTED18]], align 4 2166 // CHECK11-NEXT: [[TMP75:%.*]] = load i32, ptr [[N_CASTED18]], align 4 2167 // CHECK11-NEXT: [[TMP76:%.*]] = mul nuw i32 [[TMP0]], 4 2168 // CHECK11-NEXT: [[TMP77:%.*]] = sext i32 [[TMP76]] to i64 2169 // CHECK11-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES22]], ptr align 4 @.offload_sizes.3, i32 24, i1 false) 2170 // CHECK11-NEXT: [[TMP78:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 2171 // CHECK11-NEXT: store i32 [[TMP75]], ptr [[TMP78]], align 4 2172 // CHECK11-NEXT: [[TMP79:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 2173 // CHECK11-NEXT: store i32 [[TMP75]], ptr [[TMP79]], align 4 2174 // CHECK11-NEXT: [[TMP80:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 0 2175 // CHECK11-NEXT: store ptr null, ptr [[TMP80]], align 4 2176 // CHECK11-NEXT: [[TMP81:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 1 2177 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP81]], align 4 2178 // CHECK11-NEXT: [[TMP82:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 1 2179 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP82]], align 4 2180 // CHECK11-NEXT: [[TMP83:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 1 2181 // CHECK11-NEXT: store ptr null, ptr [[TMP83]], align 4 2182 // CHECK11-NEXT: [[TMP84:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 2 2183 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP84]], align 4 2184 // CHECK11-NEXT: [[TMP85:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 2 2185 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP85]], align 4 2186 // CHECK11-NEXT: [[TMP86:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES22]], i32 0, i32 2 2187 // CHECK11-NEXT: store i64 [[TMP77]], ptr [[TMP86]], align 4 2188 // CHECK11-NEXT: [[TMP87:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS21]], i32 0, i32 2 2189 // CHECK11-NEXT: store ptr null, ptr [[TMP87]], align 4 2190 // CHECK11-NEXT: [[TMP88:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS19]], i32 0, i32 0 2191 // CHECK11-NEXT: [[TMP89:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS20]], i32 0, i32 0 2192 // CHECK11-NEXT: [[TMP90:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES22]], i32 0, i32 0 2193 // CHECK11-NEXT: [[TMP91:%.*]] = load i32, ptr [[N]], align 4 2194 // CHECK11-NEXT: store i32 [[TMP91]], ptr [[DOTCAPTURE_EXPR_24]], align 4 2195 // CHECK11-NEXT: [[TMP92:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_24]], align 4 2196 // CHECK11-NEXT: [[SUB26:%.*]] = sub nsw i32 [[TMP92]], 0 2197 // CHECK11-NEXT: [[DIV27:%.*]] = sdiv i32 [[SUB26]], 1 2198 // CHECK11-NEXT: [[SUB28:%.*]] = sub nsw i32 [[DIV27]], 1 2199 // CHECK11-NEXT: store i32 [[SUB28]], ptr [[DOTCAPTURE_EXPR_25]], align 4 2200 // CHECK11-NEXT: [[TMP93:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_25]], align 4 2201 // CHECK11-NEXT: [[ADD29:%.*]] = add nsw i32 [[TMP93]], 1 2202 // CHECK11-NEXT: [[TMP94:%.*]] = zext i32 [[ADD29]] to i64 2203 // CHECK11-NEXT: [[TMP95:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 0 2204 // CHECK11-NEXT: store i32 3, ptr [[TMP95]], align 4 2205 // CHECK11-NEXT: [[TMP96:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 1 2206 // CHECK11-NEXT: store i32 3, ptr [[TMP96]], align 4 2207 // CHECK11-NEXT: [[TMP97:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 2 2208 // CHECK11-NEXT: store ptr [[TMP88]], ptr [[TMP97]], align 4 2209 // CHECK11-NEXT: [[TMP98:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 3 2210 // CHECK11-NEXT: store ptr [[TMP89]], ptr [[TMP98]], align 4 2211 // CHECK11-NEXT: [[TMP99:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 4 2212 // CHECK11-NEXT: store ptr [[TMP90]], ptr [[TMP99]], align 4 2213 // CHECK11-NEXT: [[TMP100:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 5 2214 // CHECK11-NEXT: store ptr @.offload_maptypes.4, ptr [[TMP100]], align 4 2215 // CHECK11-NEXT: [[TMP101:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 6 2216 // CHECK11-NEXT: store ptr null, ptr [[TMP101]], align 4 2217 // CHECK11-NEXT: [[TMP102:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 7 2218 // CHECK11-NEXT: store ptr null, ptr [[TMP102]], align 4 2219 // CHECK11-NEXT: [[TMP103:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 8 2220 // CHECK11-NEXT: store i64 [[TMP94]], ptr [[TMP103]], align 8 2221 // CHECK11-NEXT: [[TMP104:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 9 2222 // CHECK11-NEXT: store i64 0, ptr [[TMP104]], align 8 2223 // CHECK11-NEXT: [[TMP105:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 10 2224 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP105]], align 4 2225 // CHECK11-NEXT: [[TMP106:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 11 2226 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP106]], align 4 2227 // CHECK11-NEXT: [[TMP107:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS30]], i32 0, i32 12 2228 // CHECK11-NEXT: store i32 0, ptr [[TMP107]], align 4 2229 // CHECK11-NEXT: [[TMP108:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.region_id, ptr [[KERNEL_ARGS30]]) 2230 // CHECK11-NEXT: [[TMP109:%.*]] = icmp ne i32 [[TMP108]], 0 2231 // CHECK11-NEXT: br i1 [[TMP109]], label [[OMP_OFFLOAD_FAILED31:%.*]], label [[OMP_OFFLOAD_CONT32:%.*]] 2232 // CHECK11: omp_offload.failed31: 2233 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110(i32 [[TMP75]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3]] 2234 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT32]] 2235 // CHECK11: omp_offload.cont32: 2236 // CHECK11-NEXT: [[TMP110:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 2237 // CHECK11-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10EEiT_(i32 noundef [[TMP110]]) 2238 // CHECK11-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 2239 // CHECK11-NEXT: [[TMP111:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4 2240 // CHECK11-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP111]]) 2241 // CHECK11-NEXT: [[TMP112:%.*]] = load i32, ptr [[RETVAL]], align 4 2242 // CHECK11-NEXT: ret i32 [[TMP112]] 2243 // 2244 // 2245 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100 2246 // CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 2247 // CHECK11-NEXT: entry: 2248 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 2249 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 2250 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 2251 // CHECK11-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 2252 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 2253 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 2254 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 2255 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 2256 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]]) 2257 // CHECK11-NEXT: ret void 2258 // 2259 // 2260 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l100.omp_outlined 2261 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 2262 // CHECK11-NEXT: entry: 2263 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 2264 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 2265 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4 2266 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 2267 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 2268 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2269 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 2270 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2271 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2272 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 2273 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2274 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2275 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2276 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2277 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 2278 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 2279 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 2280 // CHECK11-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4 2281 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 2282 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 2283 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4 2284 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 2285 // CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4 2286 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 2287 // CHECK11-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 2288 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 2289 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 2290 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2291 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 2292 // CHECK11-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 2293 // CHECK11-NEXT: store i32 0, ptr [[I]], align 4 2294 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 2295 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 2296 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2297 // CHECK11: omp.precond.then: 2298 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2299 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2300 // CHECK11-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 2301 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2302 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2303 // CHECK11-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 2304 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 2305 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2306 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2307 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2308 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 2309 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2310 // CHECK11: cond.true: 2311 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2312 // CHECK11-NEXT: br label [[COND_END:%.*]] 2313 // CHECK11: cond.false: 2314 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2315 // CHECK11-NEXT: br label [[COND_END]] 2316 // CHECK11: cond.end: 2317 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 2318 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2319 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2320 // CHECK11-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 2321 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2322 // CHECK11: omp.inner.for.cond: 2323 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2324 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2325 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 2326 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2327 // CHECK11: omp.inner.for.body: 2328 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2329 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 2330 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2331 // CHECK11-NEXT: store i32 [[ADD]], ptr [[I3]], align 4 2332 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, ptr [[I3]], align 4 2333 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP17]] 2334 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 2335 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2336 // CHECK11: omp.body.continue: 2337 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2338 // CHECK11: omp.inner.for.inc: 2339 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2340 // CHECK11-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], 1 2341 // CHECK11-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 2342 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 2343 // CHECK11: omp.inner.for.end: 2344 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2345 // CHECK11: omp.loop.exit: 2346 // CHECK11-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 2347 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4 2348 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP20]]) 2349 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 2350 // CHECK11: omp.precond.end: 2351 // CHECK11-NEXT: ret void 2352 // 2353 // 2354 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105 2355 // CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 2356 // CHECK11-NEXT: entry: 2357 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 2358 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 2359 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 2360 // CHECK11-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 2361 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 2362 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 2363 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 2364 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 2365 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]]) 2366 // CHECK11-NEXT: ret void 2367 // 2368 // 2369 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l105.omp_outlined 2370 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 2371 // CHECK11-NEXT: entry: 2372 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 2373 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 2374 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4 2375 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 2376 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 2377 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2378 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 2379 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2380 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2381 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 2382 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2383 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2384 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2385 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2386 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 2387 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 2388 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 2389 // CHECK11-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4 2390 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 2391 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 2392 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4 2393 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 2394 // CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4 2395 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 2396 // CHECK11-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 2397 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 2398 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 2399 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2400 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 2401 // CHECK11-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 2402 // CHECK11-NEXT: store i32 0, ptr [[I]], align 4 2403 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 2404 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 2405 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2406 // CHECK11: omp.precond.then: 2407 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2408 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2409 // CHECK11-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 2410 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2411 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2412 // CHECK11-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 2413 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 2414 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2415 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2416 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2417 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 2418 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2419 // CHECK11: cond.true: 2420 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2421 // CHECK11-NEXT: br label [[COND_END:%.*]] 2422 // CHECK11: cond.false: 2423 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2424 // CHECK11-NEXT: br label [[COND_END]] 2425 // CHECK11: cond.end: 2426 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 2427 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2428 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2429 // CHECK11-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 2430 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2431 // CHECK11: omp.inner.for.cond: 2432 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2433 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2434 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 2435 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2436 // CHECK11: omp.inner.for.body: 2437 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2438 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 2439 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2440 // CHECK11-NEXT: store i32 [[ADD]], ptr [[I3]], align 4 2441 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, ptr [[I3]], align 4 2442 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP17]] 2443 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 2444 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2445 // CHECK11: omp.body.continue: 2446 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2447 // CHECK11: omp.inner.for.inc: 2448 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2449 // CHECK11-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], 1 2450 // CHECK11-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 2451 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 2452 // CHECK11: omp.inner.for.end: 2453 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2454 // CHECK11: omp.loop.exit: 2455 // CHECK11-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 2456 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4 2457 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP20]]) 2458 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 2459 // CHECK11: omp.precond.end: 2460 // CHECK11-NEXT: ret void 2461 // 2462 // 2463 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110 2464 // CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 2465 // CHECK11-NEXT: entry: 2466 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 2467 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 2468 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 2469 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2470 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 2471 // CHECK11-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 2472 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 2473 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 2474 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 2475 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 2476 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[N_ADDR]], align 4 2477 // CHECK11-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 2478 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 2479 // CHECK11-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 2480 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 2481 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]], i32 [[TMP4]]) 2482 // CHECK11-NEXT: ret void 2483 // 2484 // 2485 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l110.omp_outlined 2486 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 2487 // CHECK11-NEXT: entry: 2488 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 2489 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 2490 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4 2491 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 2492 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 2493 // CHECK11-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 2494 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2495 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 2496 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2497 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 2498 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 2499 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2500 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2501 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2502 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2503 // CHECK11-NEXT: [[I4:%.*]] = alloca i32, align 4 2504 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 2505 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 2506 // CHECK11-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4 2507 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 2508 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 2509 // CHECK11-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 2510 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4 2511 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 2512 // CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4 2513 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 2514 // CHECK11-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4 2515 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2516 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 2517 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2518 // CHECK11-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 2519 // CHECK11-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4 2520 // CHECK11-NEXT: store i32 0, ptr [[I]], align 4 2521 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2522 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 2523 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2524 // CHECK11: omp.precond.then: 2525 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2526 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 2527 // CHECK11-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 2528 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2529 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2530 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 2531 // CHECK11-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 2532 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4 2533 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP9]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 [[TMP7]]) 2534 // CHECK11-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 2535 // CHECK11: omp.dispatch.cond: 2536 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2537 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 2538 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 2539 // CHECK11-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2540 // CHECK11: cond.true: 2541 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 2542 // CHECK11-NEXT: br label [[COND_END:%.*]] 2543 // CHECK11: cond.false: 2544 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2545 // CHECK11-NEXT: br label [[COND_END]] 2546 // CHECK11: cond.end: 2547 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 2548 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2549 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2550 // CHECK11-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4 2551 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2552 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2553 // CHECK11-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 2554 // CHECK11-NEXT: br i1 [[CMP6]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 2555 // CHECK11: omp.dispatch.body: 2556 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2557 // CHECK11: omp.inner.for.cond: 2558 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12:![0-9]+]] 2559 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP12]] 2560 // CHECK11-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]] 2561 // CHECK11-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2562 // CHECK11: omp.inner.for.body: 2563 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]] 2564 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1 2565 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2566 // CHECK11-NEXT: store i32 [[ADD]], ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP12]] 2567 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP12]] 2568 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP20]] 2569 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP12]] 2570 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2571 // CHECK11: omp.body.continue: 2572 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2573 // CHECK11: omp.inner.for.inc: 2574 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]] 2575 // CHECK11-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP21]], 1 2576 // CHECK11-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]] 2577 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP13:![0-9]+]] 2578 // CHECK11: omp.inner.for.end: 2579 // CHECK11-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 2580 // CHECK11: omp.dispatch.inc: 2581 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2582 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 2583 // CHECK11-NEXT: [[ADD9:%.*]] = add nsw i32 [[TMP22]], [[TMP23]] 2584 // CHECK11-NEXT: store i32 [[ADD9]], ptr [[DOTOMP_LB]], align 4 2585 // CHECK11-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2586 // CHECK11-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 2587 // CHECK11-NEXT: [[ADD10:%.*]] = add nsw i32 [[TMP24]], [[TMP25]] 2588 // CHECK11-NEXT: store i32 [[ADD10]], ptr [[DOTOMP_UB]], align 4 2589 // CHECK11-NEXT: br label [[OMP_DISPATCH_COND]] 2590 // CHECK11: omp.dispatch.end: 2591 // CHECK11-NEXT: [[TMP26:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 2592 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, ptr [[TMP26]], align 4 2593 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP27]]) 2594 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 2595 // CHECK11: omp.precond.end: 2596 // CHECK11-NEXT: ret void 2597 // 2598 // 2599 // CHECK11-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 2600 // CHECK11-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 2601 // CHECK11-NEXT: entry: 2602 // CHECK11-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 2603 // CHECK11-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 2604 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 2605 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 2606 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 2607 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 2608 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 2609 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS1:%.*]] = alloca [1 x ptr], align 4 2610 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS2:%.*]] = alloca [1 x ptr], align 4 2611 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS3:%.*]] = alloca [1 x ptr], align 4 2612 // CHECK11-NEXT: [[_TMP4:%.*]] = alloca i32, align 4 2613 // CHECK11-NEXT: [[KERNEL_ARGS5:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 2614 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS8:%.*]] = alloca [1 x ptr], align 4 2615 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS9:%.*]] = alloca [1 x ptr], align 4 2616 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS10:%.*]] = alloca [1 x ptr], align 4 2617 // CHECK11-NEXT: [[_TMP11:%.*]] = alloca i32, align 4 2618 // CHECK11-NEXT: [[KERNEL_ARGS12:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 2619 // CHECK11-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 2620 // CHECK11-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2621 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP0]], align 4 2622 // CHECK11-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2623 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP1]], align 4 2624 // CHECK11-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 2625 // CHECK11-NEXT: store ptr null, ptr [[TMP2]], align 4 2626 // CHECK11-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2627 // CHECK11-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2628 // CHECK11-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 2629 // CHECK11-NEXT: store i32 3, ptr [[TMP5]], align 4 2630 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 2631 // CHECK11-NEXT: store i32 1, ptr [[TMP6]], align 4 2632 // CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 2633 // CHECK11-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4 2634 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 2635 // CHECK11-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4 2636 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 2637 // CHECK11-NEXT: store ptr @.offload_sizes.5, ptr [[TMP9]], align 4 2638 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 2639 // CHECK11-NEXT: store ptr @.offload_maptypes.6, ptr [[TMP10]], align 4 2640 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 2641 // CHECK11-NEXT: store ptr null, ptr [[TMP11]], align 4 2642 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 2643 // CHECK11-NEXT: store ptr null, ptr [[TMP12]], align 4 2644 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 2645 // CHECK11-NEXT: store i64 10, ptr [[TMP13]], align 8 2646 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 2647 // CHECK11-NEXT: store i64 0, ptr [[TMP14]], align 8 2648 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 2649 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4 2650 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 2651 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4 2652 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 2653 // CHECK11-NEXT: store i32 0, ptr [[TMP17]], align 4 2654 // CHECK11-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.region_id, ptr [[KERNEL_ARGS]]) 2655 // CHECK11-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0 2656 // CHECK11-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2657 // CHECK11: omp_offload.failed: 2658 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79(ptr [[A]]) #[[ATTR3]] 2659 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] 2660 // CHECK11: omp_offload.cont: 2661 // CHECK11-NEXT: [[TMP20:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 2662 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP20]], align 4 2663 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 2664 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP21]], align 4 2665 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS3]], i32 0, i32 0 2666 // CHECK11-NEXT: store ptr null, ptr [[TMP22]], align 4 2667 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS1]], i32 0, i32 0 2668 // CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS2]], i32 0, i32 0 2669 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 0 2670 // CHECK11-NEXT: store i32 3, ptr [[TMP25]], align 4 2671 // CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 1 2672 // CHECK11-NEXT: store i32 1, ptr [[TMP26]], align 4 2673 // CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 2 2674 // CHECK11-NEXT: store ptr [[TMP23]], ptr [[TMP27]], align 4 2675 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 3 2676 // CHECK11-NEXT: store ptr [[TMP24]], ptr [[TMP28]], align 4 2677 // CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 4 2678 // CHECK11-NEXT: store ptr @.offload_sizes.7, ptr [[TMP29]], align 4 2679 // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 5 2680 // CHECK11-NEXT: store ptr @.offload_maptypes.8, ptr [[TMP30]], align 4 2681 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 6 2682 // CHECK11-NEXT: store ptr null, ptr [[TMP31]], align 4 2683 // CHECK11-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 7 2684 // CHECK11-NEXT: store ptr null, ptr [[TMP32]], align 4 2685 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 8 2686 // CHECK11-NEXT: store i64 10, ptr [[TMP33]], align 8 2687 // CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 9 2688 // CHECK11-NEXT: store i64 0, ptr [[TMP34]], align 8 2689 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 10 2690 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP35]], align 4 2691 // CHECK11-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 11 2692 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP36]], align 4 2693 // CHECK11-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS5]], i32 0, i32 12 2694 // CHECK11-NEXT: store i32 0, ptr [[TMP37]], align 4 2695 // CHECK11-NEXT: [[TMP38:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.region_id, ptr [[KERNEL_ARGS5]]) 2696 // CHECK11-NEXT: [[TMP39:%.*]] = icmp ne i32 [[TMP38]], 0 2697 // CHECK11-NEXT: br i1 [[TMP39]], label [[OMP_OFFLOAD_FAILED6:%.*]], label [[OMP_OFFLOAD_CONT7:%.*]] 2698 // CHECK11: omp_offload.failed6: 2699 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84(ptr [[A]]) #[[ATTR3]] 2700 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT7]] 2701 // CHECK11: omp_offload.cont7: 2702 // CHECK11-NEXT: [[TMP40:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0 2703 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP40]], align 4 2704 // CHECK11-NEXT: [[TMP41:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0 2705 // CHECK11-NEXT: store ptr [[A]], ptr [[TMP41]], align 4 2706 // CHECK11-NEXT: [[TMP42:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS10]], i32 0, i32 0 2707 // CHECK11-NEXT: store ptr null, ptr [[TMP42]], align 4 2708 // CHECK11-NEXT: [[TMP43:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0 2709 // CHECK11-NEXT: [[TMP44:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS9]], i32 0, i32 0 2710 // CHECK11-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 0 2711 // CHECK11-NEXT: store i32 3, ptr [[TMP45]], align 4 2712 // CHECK11-NEXT: [[TMP46:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 1 2713 // CHECK11-NEXT: store i32 1, ptr [[TMP46]], align 4 2714 // CHECK11-NEXT: [[TMP47:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 2 2715 // CHECK11-NEXT: store ptr [[TMP43]], ptr [[TMP47]], align 4 2716 // CHECK11-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 3 2717 // CHECK11-NEXT: store ptr [[TMP44]], ptr [[TMP48]], align 4 2718 // CHECK11-NEXT: [[TMP49:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 4 2719 // CHECK11-NEXT: store ptr @.offload_sizes.9, ptr [[TMP49]], align 4 2720 // CHECK11-NEXT: [[TMP50:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 5 2721 // CHECK11-NEXT: store ptr @.offload_maptypes.10, ptr [[TMP50]], align 4 2722 // CHECK11-NEXT: [[TMP51:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 6 2723 // CHECK11-NEXT: store ptr null, ptr [[TMP51]], align 4 2724 // CHECK11-NEXT: [[TMP52:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 7 2725 // CHECK11-NEXT: store ptr null, ptr [[TMP52]], align 4 2726 // CHECK11-NEXT: [[TMP53:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 8 2727 // CHECK11-NEXT: store i64 10, ptr [[TMP53]], align 8 2728 // CHECK11-NEXT: [[TMP54:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 9 2729 // CHECK11-NEXT: store i64 0, ptr [[TMP54]], align 8 2730 // CHECK11-NEXT: [[TMP55:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 10 2731 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP55]], align 4 2732 // CHECK11-NEXT: [[TMP56:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 11 2733 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP56]], align 4 2734 // CHECK11-NEXT: [[TMP57:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS12]], i32 0, i32 12 2735 // CHECK11-NEXT: store i32 0, ptr [[TMP57]], align 4 2736 // CHECK11-NEXT: [[TMP58:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.region_id, ptr [[KERNEL_ARGS12]]) 2737 // CHECK11-NEXT: [[TMP59:%.*]] = icmp ne i32 [[TMP58]], 0 2738 // CHECK11-NEXT: br i1 [[TMP59]], label [[OMP_OFFLOAD_FAILED13:%.*]], label [[OMP_OFFLOAD_CONT14:%.*]] 2739 // CHECK11: omp_offload.failed13: 2740 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89(ptr [[A]]) #[[ATTR3]] 2741 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT14]] 2742 // CHECK11: omp_offload.cont14: 2743 // CHECK11-NEXT: ret i32 0 2744 // 2745 // 2746 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79 2747 // CHECK11-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2748 // CHECK11-NEXT: entry: 2749 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 2750 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 2751 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 2752 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.omp_outlined, ptr [[TMP0]]) 2753 // CHECK11-NEXT: ret void 2754 // 2755 // 2756 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l79.omp_outlined 2757 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2758 // CHECK11-NEXT: entry: 2759 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 2760 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 2761 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 2762 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2763 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 2764 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2765 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2766 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2767 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2768 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 2769 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 2770 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 2771 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 2772 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 2773 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2774 // CHECK11-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 2775 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2776 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2777 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 2778 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 2779 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2780 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2781 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 2782 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2783 // CHECK11: cond.true: 2784 // CHECK11-NEXT: br label [[COND_END:%.*]] 2785 // CHECK11: cond.false: 2786 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2787 // CHECK11-NEXT: br label [[COND_END]] 2788 // CHECK11: cond.end: 2789 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2790 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2791 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2792 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 2793 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2794 // CHECK11: omp.inner.for.cond: 2795 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2796 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2797 // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 2798 // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2799 // CHECK11: omp.inner.for.body: 2800 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2801 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 2802 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2803 // CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4 2804 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4 2805 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP9]] 2806 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 2807 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2808 // CHECK11: omp.body.continue: 2809 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2810 // CHECK11: omp.inner.for.inc: 2811 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2812 // CHECK11-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 2813 // CHECK11-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 2814 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 2815 // CHECK11: omp.inner.for.end: 2816 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2817 // CHECK11: omp.loop.exit: 2818 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 2819 // CHECK11-NEXT: ret void 2820 // 2821 // 2822 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84 2823 // CHECK11-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2824 // CHECK11-NEXT: entry: 2825 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 2826 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 2827 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 2828 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.omp_outlined, ptr [[TMP0]]) 2829 // CHECK11-NEXT: ret void 2830 // 2831 // 2832 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l84.omp_outlined 2833 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2834 // CHECK11-NEXT: entry: 2835 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 2836 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 2837 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 2838 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2839 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 2840 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2841 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2842 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2843 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2844 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 2845 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 2846 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 2847 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 2848 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 2849 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2850 // CHECK11-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 2851 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2852 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2853 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 2854 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 2855 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2856 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2857 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 2858 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2859 // CHECK11: cond.true: 2860 // CHECK11-NEXT: br label [[COND_END:%.*]] 2861 // CHECK11: cond.false: 2862 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2863 // CHECK11-NEXT: br label [[COND_END]] 2864 // CHECK11: cond.end: 2865 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2866 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2867 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2868 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 2869 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2870 // CHECK11: omp.inner.for.cond: 2871 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2872 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2873 // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 2874 // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2875 // CHECK11: omp.inner.for.body: 2876 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2877 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 2878 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2879 // CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4 2880 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4 2881 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP9]] 2882 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 2883 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2884 // CHECK11: omp.body.continue: 2885 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2886 // CHECK11: omp.inner.for.inc: 2887 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2888 // CHECK11-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 2889 // CHECK11-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 2890 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 2891 // CHECK11: omp.inner.for.end: 2892 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2893 // CHECK11: omp.loop.exit: 2894 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 2895 // CHECK11-NEXT: ret void 2896 // 2897 // 2898 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89 2899 // CHECK11-SAME: (ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2900 // CHECK11-NEXT: entry: 2901 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 2902 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 2903 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 2904 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined, ptr [[TMP0]]) 2905 // CHECK11-NEXT: ret void 2906 // 2907 // 2908 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l89.omp_outlined 2909 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2910 // CHECK11-NEXT: entry: 2911 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 2912 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 2913 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 2914 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2915 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 2916 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2917 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2918 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2919 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2920 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 2921 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 2922 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 2923 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 2924 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 2925 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2926 // CHECK11-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 2927 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2928 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2929 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 2930 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 2931 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 91, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 10) 2932 // CHECK11-NEXT: br label [[OMP_DISPATCH_COND:%.*]] 2933 // CHECK11: omp.dispatch.cond: 2934 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2935 // CHECK11-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 2936 // CHECK11-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2937 // CHECK11: cond.true: 2938 // CHECK11-NEXT: br label [[COND_END:%.*]] 2939 // CHECK11: cond.false: 2940 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2941 // CHECK11-NEXT: br label [[COND_END]] 2942 // CHECK11: cond.end: 2943 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2944 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2945 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2946 // CHECK11-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 2947 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2948 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2949 // CHECK11-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 2950 // CHECK11-NEXT: br i1 [[CMP1]], label [[OMP_DISPATCH_BODY:%.*]], label [[OMP_DISPATCH_END:%.*]] 2951 // CHECK11: omp.dispatch.body: 2952 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2953 // CHECK11: omp.inner.for.cond: 2954 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15:![0-9]+]] 2955 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP15]] 2956 // CHECK11-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 2957 // CHECK11-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2958 // CHECK11: omp.inner.for.body: 2959 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15]] 2960 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 2961 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2962 // CHECK11-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP15]] 2963 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP15]] 2964 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP11]] 2965 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP15]] 2966 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2967 // CHECK11: omp.body.continue: 2968 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2969 // CHECK11: omp.inner.for.inc: 2970 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15]] 2971 // CHECK11-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1 2972 // CHECK11-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP15]] 2973 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP16:![0-9]+]] 2974 // CHECK11: omp.inner.for.end: 2975 // CHECK11-NEXT: br label [[OMP_DISPATCH_INC:%.*]] 2976 // CHECK11: omp.dispatch.inc: 2977 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2978 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 2979 // CHECK11-NEXT: [[ADD4:%.*]] = add nsw i32 [[TMP13]], [[TMP14]] 2980 // CHECK11-NEXT: store i32 [[ADD4]], ptr [[DOTOMP_LB]], align 4 2981 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2982 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 2983 // CHECK11-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP15]], [[TMP16]] 2984 // CHECK11-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_UB]], align 4 2985 // CHECK11-NEXT: br label [[OMP_DISPATCH_COND]] 2986 // CHECK11: omp.dispatch.end: 2987 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 2988 // CHECK11-NEXT: ret void 2989 // 2990