1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _ 2 // expected-no-diagnostics 3 #ifndef HEADER 4 #define HEADER 5 // Test host codegen. 6 // RUN: %clang_cc1 -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 7 // 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 8 // 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 9 // 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 10 // 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 11 // 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 12 13 // 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}}" 14 // 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 15 // 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}}" 16 // 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}}" 17 // 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 18 // 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}}" 19 #ifdef CK1 20 21 int a[100]; 22 23 int teams_argument_global(int n){ 24 int te = n / 128; 25 int th = 128; 26 // discard n_addr 27 28 #pragma omp target 29 #pragma omp teams distribute parallel for num_teams(te), thread_limit(th) 30 for(int i = 0; i < n; i++) { 31 a[i] = 0; 32 #pragma omp cancel for 33 } 34 35 #pragma omp target 36 {{{ 37 #pragma omp teams distribute parallel for 38 for(int i = 0; i < n; i++) { 39 a[i] = 0; 40 } 41 }}} 42 43 // outlined target regions 44 45 46 47 48 return a[0]; 49 } 50 51 #endif // CK1 52 53 // Test host codegen. 54 // RUN: %clang_cc1 -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 55 // 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 56 // 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 57 // 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 58 // 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 59 // 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 60 61 // 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}}" 62 // 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 63 // 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}}" 64 // 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}}" 65 // 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 66 // 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}}" 67 #ifdef CK2 68 69 int teams_local_arg(void) { 70 int n = 100; 71 int a[n]; 72 73 #pragma omp target 74 #pragma omp teams distribute parallel for 75 for(int i = 0; i < n; i++) { 76 a[i] = 0; 77 } 78 79 // outlined target region 80 81 82 return a[0]; 83 } 84 #endif // CK2 85 86 // Test host codegen. 87 // RUN: %clang_cc1 -DCK3 -verify -Wno-vla -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK17 88 // RUN: %clang_cc1 -DCK3 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 89 // RUN: %clang_cc1 -DCK3 -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=CHECK17 90 // RUN: %clang_cc1 -DCK3 -verify -Wno-vla -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK19 91 // RUN: %clang_cc1 -DCK3 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 92 // RUN: %clang_cc1 -DCK3 -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=CHECK19 93 94 // RUN: %clang_cc1 -DCK3 -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}}" 95 // RUN: %clang_cc1 -DCK3 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 96 // RUN: %clang_cc1 -DCK3 -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}}" 97 // RUN: %clang_cc1 -DCK3 -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}}" 98 // RUN: %clang_cc1 -DCK3 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 99 // RUN: %clang_cc1 -DCK3 -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}}" 100 #ifdef CK3 101 102 103 template <typename T, int X, long long Y> 104 struct SS{ 105 T a[X]; 106 float b; 107 int foo(void) { 108 109 #pragma omp target 110 #pragma omp teams distribute parallel for 111 for(int i = 0; i < X; i++) { 112 a[i] = (T)0; 113 } 114 115 // outlined target region 116 117 118 return a[0]; 119 } 120 }; 121 122 int teams_template_struct(void) { 123 SS<int, 123, 456> V; 124 return V.foo(); 125 126 } 127 #endif // CK3 128 129 // Test host codegen. 130 // RUN: %clang_cc1 -DCK4 -verify -Wno-vla -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK25 131 // RUN: %clang_cc1 -DCK4 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 132 // RUN: %clang_cc1 -DCK4 -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=CHECK25 133 // RUN: %clang_cc1 -DCK4 -verify -Wno-vla -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK27 134 // RUN: %clang_cc1 -DCK4 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 135 // RUN: %clang_cc1 -DCK4 -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=CHECK27 136 137 // RUN: %clang_cc1 -DCK4 -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}}" 138 // RUN: %clang_cc1 -DCK4 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 139 // RUN: %clang_cc1 -DCK4 -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}}" 140 // RUN: %clang_cc1 -DCK4 -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}}" 141 // RUN: %clang_cc1 -DCK4 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 142 // RUN: %clang_cc1 -DCK4 -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}}" 143 144 #ifdef CK4 145 146 template <typename T, int n> 147 int tmain(T argc) { 148 T a[n]; 149 int te = n/128; 150 int th = 128; 151 #pragma omp target 152 #pragma omp teams distribute parallel for num_teams(te) thread_limit(th) 153 for(int i = 0; i < n; i++) { 154 a[i] = (T)0; 155 } 156 return 0; 157 } 158 159 int main (int argc, char **argv) { 160 int n = 100; 161 int a[n]; 162 #pragma omp target 163 #pragma omp teams distribute parallel for 164 for(int i = 0; i < n; i++) { 165 a[i] = 0; 166 } 167 return tmain<int, 10>(argc); 168 } 169 170 171 172 173 174 175 176 #endif // CK4 177 #endif 178 // CHECK1-LABEL: define {{[^@]+}}@_Z21teams_argument_globali 179 // CHECK1-SAME: (i32 noundef signext [[N:%.*]]) #[[ATTR0:[0-9]+]] { 180 // CHECK1-NEXT: entry: 181 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 182 // CHECK1-NEXT: [[TE:%.*]] = alloca i32, align 4 183 // CHECK1-NEXT: [[TH:%.*]] = alloca i32, align 4 184 // CHECK1-NEXT: [[TE_CASTED:%.*]] = alloca i64, align 8 185 // CHECK1-NEXT: [[TH_CASTED:%.*]] = alloca i64, align 8 186 // CHECK1-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 187 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x ptr], align 8 188 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x ptr], align 8 189 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x ptr], align 8 190 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 191 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 192 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 193 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 194 // CHECK1-NEXT: [[N_CASTED4:%.*]] = alloca i64, align 8 195 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [2 x ptr], align 8 196 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [2 x ptr], align 8 197 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [2 x ptr], align 8 198 // CHECK1-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 199 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 200 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 201 // CHECK1-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 202 // CHECK1-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 203 // CHECK1-NEXT: [[TMP0:%.*]] = load i32, ptr [[N_ADDR]], align 4 204 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP0]], 128 205 // CHECK1-NEXT: store i32 [[DIV]], ptr [[TE]], align 4 206 // CHECK1-NEXT: store i32 128, ptr [[TH]], align 4 207 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TE]], align 4 208 // CHECK1-NEXT: store i32 [[TMP1]], ptr [[TE_CASTED]], align 4 209 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, ptr [[TE_CASTED]], align 8 210 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[TH]], align 4 211 // CHECK1-NEXT: store i32 [[TMP3]], ptr [[TH_CASTED]], align 4 212 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[TH_CASTED]], align 8 213 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[N_ADDR]], align 4 214 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[N_CASTED]], align 4 215 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[N_CASTED]], align 8 216 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 217 // CHECK1-NEXT: store i64 [[TMP2]], ptr [[TMP7]], align 8 218 // CHECK1-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 219 // CHECK1-NEXT: store i64 [[TMP2]], ptr [[TMP8]], align 8 220 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 221 // CHECK1-NEXT: store ptr null, ptr [[TMP9]], align 8 222 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 223 // CHECK1-NEXT: store i64 [[TMP4]], ptr [[TMP10]], align 8 224 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 225 // CHECK1-NEXT: store i64 [[TMP4]], ptr [[TMP11]], align 8 226 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 227 // CHECK1-NEXT: store ptr null, ptr [[TMP12]], align 8 228 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 229 // CHECK1-NEXT: store i64 [[TMP6]], ptr [[TMP13]], align 8 230 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 231 // CHECK1-NEXT: store i64 [[TMP6]], ptr [[TMP14]], align 8 232 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 233 // CHECK1-NEXT: store ptr null, ptr [[TMP15]], align 8 234 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 235 // CHECK1-NEXT: store ptr @a, ptr [[TMP16]], align 8 236 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3 237 // CHECK1-NEXT: store ptr @a, ptr [[TMP17]], align 8 238 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 239 // CHECK1-NEXT: store ptr null, ptr [[TMP18]], align 8 240 // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 241 // CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 242 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[TE]], align 4 243 // CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[TH]], align 4 244 // CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[N_ADDR]], align 4 245 // CHECK1-NEXT: store i32 [[TMP23]], ptr [[DOTCAPTURE_EXPR_]], align 4 246 // CHECK1-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 247 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0 248 // CHECK1-NEXT: [[DIV2:%.*]] = sdiv i32 [[SUB]], 1 249 // CHECK1-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV2]], 1 250 // CHECK1-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_1]], align 4 251 // CHECK1-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 252 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1 253 // CHECK1-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64 254 // CHECK1-NEXT: [[TMP27:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP21]], 0 255 // CHECK1-NEXT: [[TMP28:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP22]], 0 256 // CHECK1-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 257 // CHECK1-NEXT: store i32 3, ptr [[TMP29]], align 4 258 // CHECK1-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 259 // CHECK1-NEXT: store i32 4, ptr [[TMP30]], align 4 260 // CHECK1-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 261 // CHECK1-NEXT: store ptr [[TMP19]], ptr [[TMP31]], align 8 262 // CHECK1-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 263 // CHECK1-NEXT: store ptr [[TMP20]], ptr [[TMP32]], align 8 264 // CHECK1-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 265 // CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP33]], align 8 266 // CHECK1-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 267 // CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP34]], align 8 268 // CHECK1-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 269 // CHECK1-NEXT: store ptr null, ptr [[TMP35]], align 8 270 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 271 // CHECK1-NEXT: store ptr null, ptr [[TMP36]], align 8 272 // CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 273 // CHECK1-NEXT: store i64 [[TMP26]], ptr [[TMP37]], align 8 274 // CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 275 // CHECK1-NEXT: store i64 0, ptr [[TMP38]], align 8 276 // CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 277 // CHECK1-NEXT: store [3 x i32] [[TMP27]], ptr [[TMP39]], align 4 278 // CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 279 // CHECK1-NEXT: store [3 x i32] [[TMP28]], ptr [[TMP40]], align 4 280 // CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 281 // CHECK1-NEXT: store i32 0, ptr [[TMP41]], align 4 282 // CHECK1-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 [[TMP21]], i32 [[TMP22]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28.region_id, ptr [[KERNEL_ARGS]]) 283 // CHECK1-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 284 // CHECK1-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 285 // CHECK1: omp_offload.failed: 286 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28(i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], ptr @a) #[[ATTR2:[0-9]+]] 287 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] 288 // CHECK1: omp_offload.cont: 289 // CHECK1-NEXT: [[TMP44:%.*]] = load i32, ptr [[N_ADDR]], align 4 290 // CHECK1-NEXT: store i32 [[TMP44]], ptr [[N_CASTED4]], align 4 291 // CHECK1-NEXT: [[TMP45:%.*]] = load i64, ptr [[N_CASTED4]], align 8 292 // CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 293 // CHECK1-NEXT: store i64 [[TMP45]], ptr [[TMP46]], align 8 294 // CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 295 // CHECK1-NEXT: store i64 [[TMP45]], ptr [[TMP47]], align 8 296 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 297 // CHECK1-NEXT: store ptr null, ptr [[TMP48]], align 8 298 // CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 299 // CHECK1-NEXT: store ptr @a, ptr [[TMP49]], align 8 300 // CHECK1-NEXT: [[TMP50:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 301 // CHECK1-NEXT: store ptr @a, ptr [[TMP50]], align 8 302 // CHECK1-NEXT: [[TMP51:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 303 // CHECK1-NEXT: store ptr null, ptr [[TMP51]], align 8 304 // CHECK1-NEXT: [[TMP52:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 305 // CHECK1-NEXT: [[TMP53:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 306 // CHECK1-NEXT: [[TMP54:%.*]] = load i32, ptr [[N_ADDR]], align 4 307 // CHECK1-NEXT: store i32 [[TMP54]], ptr [[DOTCAPTURE_EXPR_9]], align 4 308 // CHECK1-NEXT: [[TMP55:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_9]], align 4 309 // CHECK1-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP55]], 0 310 // CHECK1-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 311 // CHECK1-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 312 // CHECK1-NEXT: store i32 [[SUB13]], ptr [[DOTCAPTURE_EXPR_10]], align 4 313 // CHECK1-NEXT: [[TMP56:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_10]], align 4 314 // CHECK1-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP56]], 1 315 // CHECK1-NEXT: [[TMP57:%.*]] = zext i32 [[ADD14]] to i64 316 // CHECK1-NEXT: [[TMP58:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0 317 // CHECK1-NEXT: store i32 3, ptr [[TMP58]], align 4 318 // CHECK1-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1 319 // CHECK1-NEXT: store i32 2, ptr [[TMP59]], align 4 320 // CHECK1-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2 321 // CHECK1-NEXT: store ptr [[TMP52]], ptr [[TMP60]], align 8 322 // CHECK1-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3 323 // CHECK1-NEXT: store ptr [[TMP53]], ptr [[TMP61]], align 8 324 // CHECK1-NEXT: [[TMP62:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4 325 // CHECK1-NEXT: store ptr @.offload_sizes.1, ptr [[TMP62]], align 8 326 // CHECK1-NEXT: [[TMP63:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5 327 // CHECK1-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP63]], align 8 328 // CHECK1-NEXT: [[TMP64:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6 329 // CHECK1-NEXT: store ptr null, ptr [[TMP64]], align 8 330 // CHECK1-NEXT: [[TMP65:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7 331 // CHECK1-NEXT: store ptr null, ptr [[TMP65]], align 8 332 // CHECK1-NEXT: [[TMP66:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8 333 // CHECK1-NEXT: store i64 [[TMP57]], ptr [[TMP66]], align 8 334 // CHECK1-NEXT: [[TMP67:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9 335 // CHECK1-NEXT: store i64 0, ptr [[TMP67]], align 8 336 // CHECK1-NEXT: [[TMP68:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10 337 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP68]], align 4 338 // CHECK1-NEXT: [[TMP69:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11 339 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP69]], align 4 340 // CHECK1-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12 341 // CHECK1-NEXT: store i32 0, ptr [[TMP70]], align 4 342 // CHECK1-NEXT: [[TMP71:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35.region_id, ptr [[KERNEL_ARGS15]]) 343 // CHECK1-NEXT: [[TMP72:%.*]] = icmp ne i32 [[TMP71]], 0 344 // CHECK1-NEXT: br i1 [[TMP72]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] 345 // CHECK1: omp_offload.failed16: 346 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35(i64 [[TMP45]], ptr @a) #[[ATTR2]] 347 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT17]] 348 // CHECK1: omp_offload.cont17: 349 // CHECK1-NEXT: [[TMP73:%.*]] = load i32, ptr @a, align 4 350 // CHECK1-NEXT: ret i32 [[TMP73]] 351 // 352 // 353 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28 354 // CHECK1-SAME: (i64 noundef [[TE:%.*]], i64 noundef [[TH:%.*]], i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1:[0-9]+]] { 355 // CHECK1-NEXT: entry: 356 // CHECK1-NEXT: [[TE_ADDR:%.*]] = alloca i64, align 8 357 // CHECK1-NEXT: [[TH_ADDR:%.*]] = alloca i64, align 8 358 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 359 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 360 // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB3]]) 361 // CHECK1-NEXT: store i64 [[TE]], ptr [[TE_ADDR]], align 8 362 // CHECK1-NEXT: store i64 [[TH]], ptr [[TH_ADDR]], align 8 363 // CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 364 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 365 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 366 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TE_ADDR]], align 4 367 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[TH_ADDR]], align 4 368 // CHECK1-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB3]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 369 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28.omp_outlined, ptr [[N_ADDR]], ptr [[TMP1]]) 370 // CHECK1-NEXT: ret void 371 // 372 // 373 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28.omp_outlined 374 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 375 // CHECK1-NEXT: entry: 376 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 377 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 378 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8 379 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 380 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 381 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 382 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 383 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 384 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 385 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 386 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 387 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 388 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 389 // CHECK1-NEXT: [[I3:%.*]] = alloca i32, align 4 390 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 391 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 392 // CHECK1-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8 393 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 394 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8 395 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 396 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP0]], align 4 397 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 398 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 399 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 400 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 401 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 402 // CHECK1-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 403 // CHECK1-NEXT: store i32 0, ptr [[I]], align 4 404 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 405 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 406 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 407 // CHECK1: omp.precond.then: 408 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 409 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 410 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_COMB_UB]], align 4 411 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 412 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 413 // CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 414 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4 415 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP7]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 416 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 417 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 418 // CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 419 // CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 420 // CHECK1: cond.true: 421 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 422 // CHECK1-NEXT: br label [[COND_END:%.*]] 423 // CHECK1: cond.false: 424 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 425 // CHECK1-NEXT: br label [[COND_END]] 426 // CHECK1: cond.end: 427 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 428 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 429 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 430 // CHECK1-NEXT: store i32 [[TMP12]], ptr [[DOTOMP_IV]], align 4 431 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 432 // CHECK1: omp.inner.for.cond: 433 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 434 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 435 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 436 // CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 437 // CHECK1: omp.inner.for.body: 438 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 439 // CHECK1-NEXT: [[TMP16:%.*]] = zext i32 [[TMP15]] to i64 440 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 441 // CHECK1-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 442 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28.omp_outlined.omp_outlined, i64 [[TMP16]], i64 [[TMP18]], ptr [[TMP0]], ptr [[TMP1]]) 443 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 444 // CHECK1: omp.inner.for.inc: 445 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 446 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 447 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 448 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 449 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 450 // CHECK1: omp.inner.for.end: 451 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 452 // CHECK1: omp.loop.exit: 453 // CHECK1-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 454 // CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4 455 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP22]]) 456 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 457 // CHECK1: omp.precond.end: 458 // CHECK1-NEXT: ret void 459 // 460 // 461 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28.omp_outlined.omp_outlined 462 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[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: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 467 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 468 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8 469 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 470 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 471 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 472 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 473 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 474 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 475 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 476 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 477 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 478 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 479 // CHECK1-NEXT: [[I4:%.*]] = alloca i32, align 4 480 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 481 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 482 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8 483 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8 484 // CHECK1-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8 485 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 486 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8 487 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 488 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP0]], align 4 489 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 490 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 491 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 492 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 493 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 494 // CHECK1-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 495 // CHECK1-NEXT: store i32 0, ptr [[I]], align 4 496 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 497 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 498 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 499 // CHECK1: omp.precond.then: 500 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 501 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 502 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_UB]], align 4 503 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8 504 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP6]] to i32 505 // CHECK1-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 506 // CHECK1-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP7]] to i32 507 // CHECK1-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4 508 // CHECK1-NEXT: store i32 [[CONV3]], ptr [[DOTOMP_UB]], align 4 509 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 510 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 511 // CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 512 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4 513 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP9]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 514 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 515 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 516 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 517 // CHECK1-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 518 // CHECK1: cond.true: 519 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 520 // CHECK1-NEXT: br label [[COND_END:%.*]] 521 // CHECK1: cond.false: 522 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 523 // CHECK1-NEXT: br label [[COND_END]] 524 // CHECK1: cond.end: 525 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 526 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 527 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 528 // CHECK1-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4 529 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 530 // CHECK1: omp.inner.for.cond: 531 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 532 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 533 // CHECK1-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 534 // CHECK1-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 535 // CHECK1: omp.inner.for.body: 536 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 537 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 538 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 539 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I4]], align 4 540 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[I4]], align 4 541 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP18]] to i64 542 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], ptr [[TMP1]], i64 0, i64 [[IDXPROM]] 543 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 544 // CHECK1-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 545 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4 546 // CHECK1-NEXT: [[TMP21:%.*]] = call i32 @__kmpc_cancel(ptr @[[GLOB3]], i32 [[TMP20]], i32 2) 547 // CHECK1-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 548 // CHECK1-NEXT: br i1 [[TMP22]], label [[DOTCANCEL_EXIT:%.*]], label [[DOTCANCEL_CONTINUE:%.*]] 549 // CHECK1: .cancel.exit: 550 // CHECK1-NEXT: br label [[CANCEL_EXIT:%.*]] 551 // CHECK1: .cancel.continue: 552 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 553 // CHECK1: omp.body.continue: 554 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 555 // CHECK1: omp.inner.for.inc: 556 // CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 557 // CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP23]], 1 558 // CHECK1-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4 559 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 560 // CHECK1: omp.inner.for.end: 561 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 562 // CHECK1: omp.loop.exit: 563 // CHECK1-NEXT: [[TMP24:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 564 // CHECK1-NEXT: [[TMP25:%.*]] = load i32, ptr [[TMP24]], align 4 565 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP25]]) 566 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 567 // CHECK1: cancel.exit: 568 // CHECK1-NEXT: [[TMP26:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 569 // CHECK1-NEXT: [[TMP27:%.*]] = load i32, ptr [[TMP26]], align 4 570 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP27]]) 571 // CHECK1-NEXT: br label [[CANCEL_CONT:%.*]] 572 // CHECK1: omp.precond.end: 573 // CHECK1-NEXT: br label [[CANCEL_CONT]] 574 // CHECK1: cancel.cont: 575 // CHECK1-NEXT: ret void 576 // 577 // 578 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35 579 // CHECK1-SAME: (i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 580 // CHECK1-NEXT: entry: 581 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 582 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 583 // CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 584 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 585 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 586 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35.omp_outlined, ptr [[N_ADDR]], ptr [[TMP0]]) 587 // CHECK1-NEXT: ret void 588 // 589 // 590 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35.omp_outlined 591 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 592 // CHECK1-NEXT: entry: 593 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 594 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 595 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8 596 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 597 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 598 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 599 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 600 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 601 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 602 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 603 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 604 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 605 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 606 // CHECK1-NEXT: [[I3:%.*]] = alloca i32, align 4 607 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 608 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 609 // CHECK1-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8 610 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 611 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8 612 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 613 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP0]], align 4 614 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 615 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 616 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 617 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 618 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 619 // CHECK1-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 620 // CHECK1-NEXT: store i32 0, ptr [[I]], align 4 621 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 622 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 623 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 624 // CHECK1: omp.precond.then: 625 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 626 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 627 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_COMB_UB]], align 4 628 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 629 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 630 // CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 631 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4 632 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP7]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 633 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 634 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 635 // CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 636 // CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 637 // CHECK1: cond.true: 638 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 639 // CHECK1-NEXT: br label [[COND_END:%.*]] 640 // CHECK1: cond.false: 641 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 642 // CHECK1-NEXT: br label [[COND_END]] 643 // CHECK1: cond.end: 644 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 645 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 646 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 647 // CHECK1-NEXT: store i32 [[TMP12]], ptr [[DOTOMP_IV]], align 4 648 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 649 // CHECK1: omp.inner.for.cond: 650 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 651 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 652 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 653 // CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 654 // CHECK1: omp.inner.for.body: 655 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 656 // CHECK1-NEXT: [[TMP16:%.*]] = zext i32 [[TMP15]] to i64 657 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 658 // CHECK1-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 659 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35.omp_outlined.omp_outlined, i64 [[TMP16]], i64 [[TMP18]], ptr [[TMP0]], ptr [[TMP1]]) 660 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 661 // CHECK1: omp.inner.for.inc: 662 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 663 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 664 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 665 // CHECK1-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 666 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 667 // CHECK1: omp.inner.for.end: 668 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 669 // CHECK1: omp.loop.exit: 670 // CHECK1-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 671 // CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4 672 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP22]]) 673 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 674 // CHECK1: omp.precond.end: 675 // CHECK1-NEXT: ret void 676 // 677 // 678 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35.omp_outlined.omp_outlined 679 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 680 // CHECK1-NEXT: entry: 681 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 682 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 683 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 684 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 685 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8 686 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 687 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 688 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 689 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 690 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 691 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 692 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 693 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 694 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 695 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 696 // CHECK1-NEXT: [[I4:%.*]] = alloca i32, align 4 697 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 698 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 699 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8 700 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8 701 // CHECK1-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8 702 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 703 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8 704 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 705 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP0]], align 4 706 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 707 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 708 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 709 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 710 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 711 // CHECK1-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 712 // CHECK1-NEXT: store i32 0, ptr [[I]], align 4 713 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 714 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 715 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 716 // CHECK1: omp.precond.then: 717 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 718 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 719 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_UB]], align 4 720 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8 721 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP6]] to i32 722 // CHECK1-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 723 // CHECK1-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP7]] to i32 724 // CHECK1-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4 725 // CHECK1-NEXT: store i32 [[CONV3]], ptr [[DOTOMP_UB]], align 4 726 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 727 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 728 // CHECK1-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 729 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4 730 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP9]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 731 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 732 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 733 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 734 // CHECK1-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 735 // CHECK1: cond.true: 736 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 737 // CHECK1-NEXT: br label [[COND_END:%.*]] 738 // CHECK1: cond.false: 739 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 740 // CHECK1-NEXT: br label [[COND_END]] 741 // CHECK1: cond.end: 742 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 743 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 744 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 745 // CHECK1-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4 746 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 747 // CHECK1: omp.inner.for.cond: 748 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 749 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 750 // CHECK1-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 751 // CHECK1-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 752 // CHECK1: omp.inner.for.body: 753 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 754 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 755 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 756 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I4]], align 4 757 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[I4]], align 4 758 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP18]] to i64 759 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], ptr [[TMP1]], i64 0, i64 [[IDXPROM]] 760 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 761 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 762 // CHECK1: omp.body.continue: 763 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 764 // CHECK1: omp.inner.for.inc: 765 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 766 // CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP19]], 1 767 // CHECK1-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4 768 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 769 // CHECK1: omp.inner.for.end: 770 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 771 // CHECK1: omp.loop.exit: 772 // CHECK1-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 773 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4 774 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP21]]) 775 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 776 // CHECK1: omp.precond.end: 777 // CHECK1-NEXT: ret void 778 // 779 // 780 // CHECK3-LABEL: define {{[^@]+}}@_Z21teams_argument_globali 781 // CHECK3-SAME: (i32 noundef [[N:%.*]]) #[[ATTR0:[0-9]+]] { 782 // CHECK3-NEXT: entry: 783 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 784 // CHECK3-NEXT: [[TE:%.*]] = alloca i32, align 4 785 // CHECK3-NEXT: [[TH:%.*]] = alloca i32, align 4 786 // CHECK3-NEXT: [[TE_CASTED:%.*]] = alloca i32, align 4 787 // CHECK3-NEXT: [[TH_CASTED:%.*]] = alloca i32, align 4 788 // CHECK3-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 789 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x ptr], align 4 790 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x ptr], align 4 791 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x ptr], align 4 792 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 793 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 794 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 795 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 796 // CHECK3-NEXT: [[N_CASTED4:%.*]] = alloca i32, align 4 797 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [2 x ptr], align 4 798 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [2 x ptr], align 4 799 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [2 x ptr], align 4 800 // CHECK3-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 801 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 802 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 803 // CHECK3-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 804 // CHECK3-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 805 // CHECK3-NEXT: [[TMP0:%.*]] = load i32, ptr [[N_ADDR]], align 4 806 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP0]], 128 807 // CHECK3-NEXT: store i32 [[DIV]], ptr [[TE]], align 4 808 // CHECK3-NEXT: store i32 128, ptr [[TH]], align 4 809 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[TE]], align 4 810 // CHECK3-NEXT: store i32 [[TMP1]], ptr [[TE_CASTED]], align 4 811 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TE_CASTED]], align 4 812 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[TH]], align 4 813 // CHECK3-NEXT: store i32 [[TMP3]], ptr [[TH_CASTED]], align 4 814 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[TH_CASTED]], align 4 815 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[N_ADDR]], align 4 816 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[N_CASTED]], align 4 817 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[N_CASTED]], align 4 818 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 819 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[TMP7]], align 4 820 // CHECK3-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 821 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[TMP8]], align 4 822 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 823 // CHECK3-NEXT: store ptr null, ptr [[TMP9]], align 4 824 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 825 // CHECK3-NEXT: store i32 [[TMP4]], ptr [[TMP10]], align 4 826 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 827 // CHECK3-NEXT: store i32 [[TMP4]], ptr [[TMP11]], align 4 828 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 829 // CHECK3-NEXT: store ptr null, ptr [[TMP12]], align 4 830 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 831 // CHECK3-NEXT: store i32 [[TMP6]], ptr [[TMP13]], align 4 832 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 833 // CHECK3-NEXT: store i32 [[TMP6]], ptr [[TMP14]], align 4 834 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 835 // CHECK3-NEXT: store ptr null, ptr [[TMP15]], align 4 836 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 837 // CHECK3-NEXT: store ptr @a, ptr [[TMP16]], align 4 838 // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3 839 // CHECK3-NEXT: store ptr @a, ptr [[TMP17]], align 4 840 // CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 841 // CHECK3-NEXT: store ptr null, ptr [[TMP18]], align 4 842 // CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 843 // CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 844 // CHECK3-NEXT: [[TMP21:%.*]] = load i32, ptr [[TE]], align 4 845 // CHECK3-NEXT: [[TMP22:%.*]] = load i32, ptr [[TH]], align 4 846 // CHECK3-NEXT: [[TMP23:%.*]] = load i32, ptr [[N_ADDR]], align 4 847 // CHECK3-NEXT: store i32 [[TMP23]], ptr [[DOTCAPTURE_EXPR_]], align 4 848 // CHECK3-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 849 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP24]], 0 850 // CHECK3-NEXT: [[DIV2:%.*]] = sdiv i32 [[SUB]], 1 851 // CHECK3-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV2]], 1 852 // CHECK3-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_1]], align 4 853 // CHECK3-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 854 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP25]], 1 855 // CHECK3-NEXT: [[TMP26:%.*]] = zext i32 [[ADD]] to i64 856 // CHECK3-NEXT: [[TMP27:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP21]], 0 857 // CHECK3-NEXT: [[TMP28:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP22]], 0 858 // CHECK3-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 859 // CHECK3-NEXT: store i32 3, ptr [[TMP29]], align 4 860 // CHECK3-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 861 // CHECK3-NEXT: store i32 4, ptr [[TMP30]], align 4 862 // CHECK3-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 863 // CHECK3-NEXT: store ptr [[TMP19]], ptr [[TMP31]], align 4 864 // CHECK3-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 865 // CHECK3-NEXT: store ptr [[TMP20]], ptr [[TMP32]], align 4 866 // CHECK3-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 867 // CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP33]], align 4 868 // CHECK3-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 869 // CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP34]], align 4 870 // CHECK3-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 871 // CHECK3-NEXT: store ptr null, ptr [[TMP35]], align 4 872 // CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 873 // CHECK3-NEXT: store ptr null, ptr [[TMP36]], align 4 874 // CHECK3-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 875 // CHECK3-NEXT: store i64 [[TMP26]], ptr [[TMP37]], align 8 876 // CHECK3-NEXT: [[TMP38:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 877 // CHECK3-NEXT: store i64 0, ptr [[TMP38]], align 8 878 // CHECK3-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 879 // CHECK3-NEXT: store [3 x i32] [[TMP27]], ptr [[TMP39]], align 4 880 // CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 881 // CHECK3-NEXT: store [3 x i32] [[TMP28]], ptr [[TMP40]], align 4 882 // CHECK3-NEXT: [[TMP41:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 883 // CHECK3-NEXT: store i32 0, ptr [[TMP41]], align 4 884 // CHECK3-NEXT: [[TMP42:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 [[TMP21]], i32 [[TMP22]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28.region_id, ptr [[KERNEL_ARGS]]) 885 // CHECK3-NEXT: [[TMP43:%.*]] = icmp ne i32 [[TMP42]], 0 886 // CHECK3-NEXT: br i1 [[TMP43]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 887 // CHECK3: omp_offload.failed: 888 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28(i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], ptr @a) #[[ATTR2:[0-9]+]] 889 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] 890 // CHECK3: omp_offload.cont: 891 // CHECK3-NEXT: [[TMP44:%.*]] = load i32, ptr [[N_ADDR]], align 4 892 // CHECK3-NEXT: store i32 [[TMP44]], ptr [[N_CASTED4]], align 4 893 // CHECK3-NEXT: [[TMP45:%.*]] = load i32, ptr [[N_CASTED4]], align 4 894 // CHECK3-NEXT: [[TMP46:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 895 // CHECK3-NEXT: store i32 [[TMP45]], ptr [[TMP46]], align 4 896 // CHECK3-NEXT: [[TMP47:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 897 // CHECK3-NEXT: store i32 [[TMP45]], ptr [[TMP47]], align 4 898 // CHECK3-NEXT: [[TMP48:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS7]], i32 0, i32 0 899 // CHECK3-NEXT: store ptr null, ptr [[TMP48]], align 4 900 // CHECK3-NEXT: [[TMP49:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 901 // CHECK3-NEXT: store ptr @a, ptr [[TMP49]], align 4 902 // CHECK3-NEXT: [[TMP50:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 903 // CHECK3-NEXT: store ptr @a, ptr [[TMP50]], align 4 904 // CHECK3-NEXT: [[TMP51:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS7]], i32 0, i32 1 905 // CHECK3-NEXT: store ptr null, ptr [[TMP51]], align 4 906 // CHECK3-NEXT: [[TMP52:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 907 // CHECK3-NEXT: [[TMP53:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 908 // CHECK3-NEXT: [[TMP54:%.*]] = load i32, ptr [[N_ADDR]], align 4 909 // CHECK3-NEXT: store i32 [[TMP54]], ptr [[DOTCAPTURE_EXPR_9]], align 4 910 // CHECK3-NEXT: [[TMP55:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_9]], align 4 911 // CHECK3-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP55]], 0 912 // CHECK3-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 913 // CHECK3-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 914 // CHECK3-NEXT: store i32 [[SUB13]], ptr [[DOTCAPTURE_EXPR_10]], align 4 915 // CHECK3-NEXT: [[TMP56:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_10]], align 4 916 // CHECK3-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP56]], 1 917 // CHECK3-NEXT: [[TMP57:%.*]] = zext i32 [[ADD14]] to i64 918 // CHECK3-NEXT: [[TMP58:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0 919 // CHECK3-NEXT: store i32 3, ptr [[TMP58]], align 4 920 // CHECK3-NEXT: [[TMP59:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1 921 // CHECK3-NEXT: store i32 2, ptr [[TMP59]], align 4 922 // CHECK3-NEXT: [[TMP60:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2 923 // CHECK3-NEXT: store ptr [[TMP52]], ptr [[TMP60]], align 4 924 // CHECK3-NEXT: [[TMP61:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3 925 // CHECK3-NEXT: store ptr [[TMP53]], ptr [[TMP61]], align 4 926 // CHECK3-NEXT: [[TMP62:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4 927 // CHECK3-NEXT: store ptr @.offload_sizes.1, ptr [[TMP62]], align 4 928 // CHECK3-NEXT: [[TMP63:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5 929 // CHECK3-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP63]], align 4 930 // CHECK3-NEXT: [[TMP64:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6 931 // CHECK3-NEXT: store ptr null, ptr [[TMP64]], align 4 932 // CHECK3-NEXT: [[TMP65:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7 933 // CHECK3-NEXT: store ptr null, ptr [[TMP65]], align 4 934 // CHECK3-NEXT: [[TMP66:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8 935 // CHECK3-NEXT: store i64 [[TMP57]], ptr [[TMP66]], align 8 936 // CHECK3-NEXT: [[TMP67:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9 937 // CHECK3-NEXT: store i64 0, ptr [[TMP67]], align 8 938 // CHECK3-NEXT: [[TMP68:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10 939 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP68]], align 4 940 // CHECK3-NEXT: [[TMP69:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11 941 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP69]], align 4 942 // CHECK3-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12 943 // CHECK3-NEXT: store i32 0, ptr [[TMP70]], align 4 944 // CHECK3-NEXT: [[TMP71:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35.region_id, ptr [[KERNEL_ARGS15]]) 945 // CHECK3-NEXT: [[TMP72:%.*]] = icmp ne i32 [[TMP71]], 0 946 // CHECK3-NEXT: br i1 [[TMP72]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] 947 // CHECK3: omp_offload.failed16: 948 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35(i32 [[TMP45]], ptr @a) #[[ATTR2]] 949 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT17]] 950 // CHECK3: omp_offload.cont17: 951 // CHECK3-NEXT: [[TMP73:%.*]] = load i32, ptr @a, align 4 952 // CHECK3-NEXT: ret i32 [[TMP73]] 953 // 954 // 955 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28 956 // CHECK3-SAME: (i32 noundef [[TE:%.*]], i32 noundef [[TH:%.*]], i32 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1:[0-9]+]] { 957 // CHECK3-NEXT: entry: 958 // CHECK3-NEXT: [[TE_ADDR:%.*]] = alloca i32, align 4 959 // CHECK3-NEXT: [[TH_ADDR:%.*]] = alloca i32, align 4 960 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 961 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 962 // CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB3]]) 963 // CHECK3-NEXT: store i32 [[TE]], ptr [[TE_ADDR]], align 4 964 // CHECK3-NEXT: store i32 [[TH]], ptr [[TH_ADDR]], align 4 965 // CHECK3-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 966 // CHECK3-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 967 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 968 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TE_ADDR]], align 4 969 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[TH_ADDR]], align 4 970 // CHECK3-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB3]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 971 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28.omp_outlined, ptr [[N_ADDR]], ptr [[TMP1]]) 972 // CHECK3-NEXT: ret void 973 // 974 // 975 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28.omp_outlined 976 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 977 // CHECK3-NEXT: entry: 978 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 979 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 980 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4 981 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 982 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 983 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 984 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 985 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 986 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 987 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 988 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 989 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 990 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 991 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 992 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 993 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 994 // CHECK3-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4 995 // CHECK3-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 996 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4 997 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 998 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP0]], align 4 999 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 1000 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1001 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 1002 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1003 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1004 // CHECK3-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1005 // CHECK3-NEXT: store i32 0, ptr [[I]], align 4 1006 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1007 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 1008 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1009 // CHECK3: omp.precond.then: 1010 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 1011 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1012 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_COMB_UB]], align 4 1013 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1014 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1015 // CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1016 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4 1017 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP7]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1018 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1019 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1020 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 1021 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1022 // CHECK3: cond.true: 1023 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1024 // CHECK3-NEXT: br label [[COND_END:%.*]] 1025 // CHECK3: cond.false: 1026 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1027 // CHECK3-NEXT: br label [[COND_END]] 1028 // CHECK3: cond.end: 1029 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 1030 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 1031 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1032 // CHECK3-NEXT: store i32 [[TMP12]], ptr [[DOTOMP_IV]], align 4 1033 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1034 // CHECK3: omp.inner.for.cond: 1035 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1036 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1037 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 1038 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1039 // CHECK3: omp.inner.for.body: 1040 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1041 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1042 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28.omp_outlined.omp_outlined, i32 [[TMP15]], i32 [[TMP16]], ptr [[TMP0]], ptr [[TMP1]]) 1043 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1044 // CHECK3: omp.inner.for.inc: 1045 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1046 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 1047 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP17]], [[TMP18]] 1048 // CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 1049 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 1050 // CHECK3: omp.inner.for.end: 1051 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1052 // CHECK3: omp.loop.exit: 1053 // CHECK3-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1054 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4 1055 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP20]]) 1056 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 1057 // CHECK3: omp.precond.end: 1058 // CHECK3-NEXT: ret void 1059 // 1060 // 1061 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28.omp_outlined.omp_outlined 1062 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 1063 // CHECK3-NEXT: entry: 1064 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 1065 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 1066 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 1067 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 1068 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4 1069 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 1070 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1071 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 1072 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1073 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1074 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 1075 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1076 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1077 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1078 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1079 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 1080 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 1081 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 1082 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4 1083 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4 1084 // CHECK3-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4 1085 // CHECK3-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 1086 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4 1087 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 1088 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP0]], align 4 1089 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 1090 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1091 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 1092 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1093 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1094 // CHECK3-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1095 // CHECK3-NEXT: store i32 0, ptr [[I]], align 4 1096 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1097 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 1098 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1099 // CHECK3: omp.precond.then: 1100 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1101 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1102 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_UB]], align 4 1103 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4 1104 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4 1105 // CHECK3-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_LB]], align 4 1106 // CHECK3-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_UB]], align 4 1107 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1108 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1109 // CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1110 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4 1111 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP9]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1112 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1113 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1114 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 1115 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1116 // CHECK3: cond.true: 1117 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1118 // CHECK3-NEXT: br label [[COND_END:%.*]] 1119 // CHECK3: cond.false: 1120 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1121 // CHECK3-NEXT: br label [[COND_END]] 1122 // CHECK3: cond.end: 1123 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 1124 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 1125 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1126 // CHECK3-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4 1127 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1128 // CHECK3: omp.inner.for.cond: 1129 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1130 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1131 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 1132 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1133 // CHECK3: omp.inner.for.body: 1134 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1135 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 1136 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1137 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I3]], align 4 1138 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[I3]], align 4 1139 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], ptr [[TMP1]], i32 0, i32 [[TMP18]] 1140 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 1141 // CHECK3-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1142 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4 1143 // CHECK3-NEXT: [[TMP21:%.*]] = call i32 @__kmpc_cancel(ptr @[[GLOB3]], i32 [[TMP20]], i32 2) 1144 // CHECK3-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 1145 // CHECK3-NEXT: br i1 [[TMP22]], label [[DOTCANCEL_EXIT:%.*]], label [[DOTCANCEL_CONTINUE:%.*]] 1146 // CHECK3: .cancel.exit: 1147 // CHECK3-NEXT: br label [[CANCEL_EXIT:%.*]] 1148 // CHECK3: .cancel.continue: 1149 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1150 // CHECK3: omp.body.continue: 1151 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1152 // CHECK3: omp.inner.for.inc: 1153 // CHECK3-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1154 // CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP23]], 1 1155 // CHECK3-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 1156 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 1157 // CHECK3: omp.inner.for.end: 1158 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1159 // CHECK3: omp.loop.exit: 1160 // CHECK3-NEXT: [[TMP24:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1161 // CHECK3-NEXT: [[TMP25:%.*]] = load i32, ptr [[TMP24]], align 4 1162 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP25]]) 1163 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 1164 // CHECK3: cancel.exit: 1165 // CHECK3-NEXT: [[TMP26:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1166 // CHECK3-NEXT: [[TMP27:%.*]] = load i32, ptr [[TMP26]], align 4 1167 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP27]]) 1168 // CHECK3-NEXT: br label [[CANCEL_CONT:%.*]] 1169 // CHECK3: omp.precond.end: 1170 // CHECK3-NEXT: br label [[CANCEL_CONT]] 1171 // CHECK3: cancel.cont: 1172 // CHECK3-NEXT: ret void 1173 // 1174 // 1175 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35 1176 // CHECK3-SAME: (i32 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 1177 // CHECK3-NEXT: entry: 1178 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 1179 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 1180 // CHECK3-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 1181 // CHECK3-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 1182 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 1183 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35.omp_outlined, ptr [[N_ADDR]], ptr [[TMP0]]) 1184 // CHECK3-NEXT: ret void 1185 // 1186 // 1187 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35.omp_outlined 1188 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 1189 // CHECK3-NEXT: entry: 1190 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 1191 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 1192 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4 1193 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 1194 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1195 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 1196 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1197 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1198 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 1199 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 1200 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 1201 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1202 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1203 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 1204 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 1205 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 1206 // CHECK3-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4 1207 // CHECK3-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 1208 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4 1209 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 1210 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP0]], align 4 1211 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 1212 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1213 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 1214 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1215 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1216 // CHECK3-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1217 // CHECK3-NEXT: store i32 0, ptr [[I]], align 4 1218 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1219 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 1220 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1221 // CHECK3: omp.precond.then: 1222 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 1223 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1224 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_COMB_UB]], align 4 1225 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1226 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1227 // CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1228 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4 1229 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP7]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1230 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1231 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1232 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 1233 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1234 // CHECK3: cond.true: 1235 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1236 // CHECK3-NEXT: br label [[COND_END:%.*]] 1237 // CHECK3: cond.false: 1238 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1239 // CHECK3-NEXT: br label [[COND_END]] 1240 // CHECK3: cond.end: 1241 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 1242 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 1243 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1244 // CHECK3-NEXT: store i32 [[TMP12]], ptr [[DOTOMP_IV]], align 4 1245 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1246 // CHECK3: omp.inner.for.cond: 1247 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1248 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1249 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 1250 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1251 // CHECK3: omp.inner.for.body: 1252 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1253 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1254 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35.omp_outlined.omp_outlined, i32 [[TMP15]], i32 [[TMP16]], ptr [[TMP0]], ptr [[TMP1]]) 1255 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1256 // CHECK3: omp.inner.for.inc: 1257 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1258 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 1259 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP17]], [[TMP18]] 1260 // CHECK3-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 1261 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 1262 // CHECK3: omp.inner.for.end: 1263 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1264 // CHECK3: omp.loop.exit: 1265 // CHECK3-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1266 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4 1267 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP20]]) 1268 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 1269 // CHECK3: omp.precond.end: 1270 // CHECK3-NEXT: ret void 1271 // 1272 // 1273 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35.omp_outlined.omp_outlined 1274 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 1275 // CHECK3-NEXT: entry: 1276 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 1277 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 1278 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 1279 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 1280 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4 1281 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 1282 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1283 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 1284 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1285 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1286 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 1287 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1288 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1289 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1290 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1291 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 1292 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 1293 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 1294 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4 1295 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4 1296 // CHECK3-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4 1297 // CHECK3-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 1298 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4 1299 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 1300 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP0]], align 4 1301 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 1302 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1303 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 1304 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1305 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1306 // CHECK3-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1307 // CHECK3-NEXT: store i32 0, ptr [[I]], align 4 1308 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1309 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 1310 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1311 // CHECK3: omp.precond.then: 1312 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1313 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1314 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_UB]], align 4 1315 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4 1316 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4 1317 // CHECK3-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_LB]], align 4 1318 // CHECK3-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_UB]], align 4 1319 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1320 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1321 // CHECK3-NEXT: [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1322 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4 1323 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP9]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1324 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1325 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1326 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 1327 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1328 // CHECK3: cond.true: 1329 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1330 // CHECK3-NEXT: br label [[COND_END:%.*]] 1331 // CHECK3: cond.false: 1332 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1333 // CHECK3-NEXT: br label [[COND_END]] 1334 // CHECK3: cond.end: 1335 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 1336 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 1337 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1338 // CHECK3-NEXT: store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4 1339 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1340 // CHECK3: omp.inner.for.cond: 1341 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1342 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1343 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 1344 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1345 // CHECK3: omp.inner.for.body: 1346 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1347 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 1348 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1349 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I3]], align 4 1350 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[I3]], align 4 1351 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], ptr [[TMP1]], i32 0, i32 [[TMP18]] 1352 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 1353 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1354 // CHECK3: omp.body.continue: 1355 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1356 // CHECK3: omp.inner.for.inc: 1357 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1358 // CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP19]], 1 1359 // CHECK3-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 1360 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 1361 // CHECK3: omp.inner.for.end: 1362 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1363 // CHECK3: omp.loop.exit: 1364 // CHECK3-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1365 // CHECK3-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4 1366 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP21]]) 1367 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 1368 // CHECK3: omp.precond.end: 1369 // CHECK3-NEXT: ret void 1370 // 1371 // 1372 // CHECK9-LABEL: define {{[^@]+}}@_Z15teams_local_argv 1373 // CHECK9-SAME: () #[[ATTR0:[0-9]+]] { 1374 // CHECK9-NEXT: entry: 1375 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 1376 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 8 1377 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 1378 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 1379 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8 1380 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8 1381 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8 1382 // CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8 1383 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1384 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1385 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1386 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 1387 // CHECK9-NEXT: store i32 100, ptr [[N]], align 4 1388 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 1389 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 1390 // CHECK9-NEXT: [[TMP2:%.*]] = call ptr @llvm.stacksave.p0() 1391 // CHECK9-NEXT: store ptr [[TMP2]], ptr [[SAVED_STACK]], align 8 1392 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 1393 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8 1394 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N]], align 4 1395 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 1396 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8 1397 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 1398 // CHECK9-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 24, i1 false) 1399 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1400 // CHECK9-NEXT: store i64 [[TMP4]], ptr [[TMP6]], align 8 1401 // CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1402 // CHECK9-NEXT: store i64 [[TMP4]], ptr [[TMP7]], align 8 1403 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 1404 // CHECK9-NEXT: store ptr null, ptr [[TMP8]], align 8 1405 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 1406 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP9]], align 8 1407 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 1408 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP10]], align 8 1409 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 1410 // CHECK9-NEXT: store ptr null, ptr [[TMP11]], align 8 1411 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 1412 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP12]], align 8 1413 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 1414 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP13]], align 8 1415 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2 1416 // CHECK9-NEXT: store i64 [[TMP5]], ptr [[TMP14]], align 8 1417 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 1418 // CHECK9-NEXT: store ptr null, ptr [[TMP15]], align 8 1419 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1420 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1421 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 1422 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, ptr [[N]], align 4 1423 // CHECK9-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4 1424 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1425 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0 1426 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1427 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1428 // CHECK9-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1429 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1430 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], 1 1431 // CHECK9-NEXT: [[TMP22:%.*]] = zext i32 [[ADD]] to i64 1432 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 1433 // CHECK9-NEXT: store i32 3, ptr [[TMP23]], align 4 1434 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 1435 // CHECK9-NEXT: store i32 3, ptr [[TMP24]], align 4 1436 // CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 1437 // CHECK9-NEXT: store ptr [[TMP16]], ptr [[TMP25]], align 8 1438 // CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 1439 // CHECK9-NEXT: store ptr [[TMP17]], ptr [[TMP26]], align 8 1440 // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 1441 // CHECK9-NEXT: store ptr [[TMP18]], ptr [[TMP27]], align 8 1442 // CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 1443 // CHECK9-NEXT: store ptr @.offload_maptypes, ptr [[TMP28]], align 8 1444 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 1445 // CHECK9-NEXT: store ptr null, ptr [[TMP29]], align 8 1446 // CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 1447 // CHECK9-NEXT: store ptr null, ptr [[TMP30]], align 8 1448 // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 1449 // CHECK9-NEXT: store i64 [[TMP22]], ptr [[TMP31]], align 8 1450 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 1451 // CHECK9-NEXT: store i64 0, ptr [[TMP32]], align 8 1452 // CHECK9-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 1453 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4 1454 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 1455 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP34]], align 4 1456 // CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 1457 // CHECK9-NEXT: store i32 0, ptr [[TMP35]], align 4 1458 // CHECK9-NEXT: [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, ptr [[KERNEL_ARGS]]) 1459 // CHECK9-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0 1460 // CHECK9-NEXT: br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1461 // CHECK9: omp_offload.failed: 1462 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i64 [[TMP4]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3:[0-9]+]] 1463 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] 1464 // CHECK9: omp_offload.cont: 1465 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i64 0 1466 // CHECK9-NEXT: [[TMP38:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 1467 // CHECK9-NEXT: [[TMP39:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 1468 // CHECK9-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP39]]) 1469 // CHECK9-NEXT: ret i32 [[TMP38]] 1470 // 1471 // 1472 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73 1473 // CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 1474 // CHECK9-NEXT: entry: 1475 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 1476 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1477 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1478 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 1479 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 1480 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1481 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 1482 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1483 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]]) 1484 // CHECK9-NEXT: ret void 1485 // 1486 // 1487 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.omp_outlined 1488 // 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]] { 1489 // CHECK9-NEXT: entry: 1490 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 1491 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 1492 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8 1493 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1494 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1495 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1496 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1497 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1498 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1499 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 1500 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 1501 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 1502 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1503 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1504 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 1505 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 1506 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 1507 // CHECK9-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8 1508 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 1509 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1510 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8 1511 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 1512 // CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1513 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 1514 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 1515 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1516 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1517 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1518 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1519 // CHECK9-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1520 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4 1521 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1522 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1523 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1524 // CHECK9: omp.precond.then: 1525 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 1526 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1527 // CHECK9-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4 1528 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1529 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1530 // CHECK9-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1531 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 1532 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1533 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1534 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1535 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 1536 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1537 // CHECK9: cond.true: 1538 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1539 // CHECK9-NEXT: br label [[COND_END:%.*]] 1540 // CHECK9: cond.false: 1541 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1542 // CHECK9-NEXT: br label [[COND_END]] 1543 // CHECK9: cond.end: 1544 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 1545 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 1546 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1547 // CHECK9-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 1548 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1549 // CHECK9: omp.inner.for.cond: 1550 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1551 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1552 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 1553 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1554 // CHECK9: omp.inner.for.body: 1555 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1556 // CHECK9-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64 1557 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1558 // CHECK9-NEXT: [[TMP19:%.*]] = zext i32 [[TMP18]] to i64 1559 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.omp_outlined.omp_outlined, i64 [[TMP17]], i64 [[TMP19]], ptr [[TMP0]], i64 [[TMP1]], ptr [[TMP2]]) 1560 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1561 // CHECK9: omp.inner.for.inc: 1562 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1563 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 1564 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]] 1565 // CHECK9-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 1566 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 1567 // CHECK9: omp.inner.for.end: 1568 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1569 // CHECK9: omp.loop.exit: 1570 // CHECK9-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1571 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4 1572 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP23]]) 1573 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 1574 // CHECK9: omp.precond.end: 1575 // CHECK9-NEXT: ret void 1576 // 1577 // 1578 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.omp_outlined.omp_outlined 1579 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1580 // CHECK9-NEXT: entry: 1581 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 1582 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 1583 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 1584 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 1585 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8 1586 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1587 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1588 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1589 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1590 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1591 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1592 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 1593 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1594 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1595 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1596 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1597 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 1598 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 1599 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 1600 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8 1601 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8 1602 // CHECK9-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8 1603 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 1604 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1605 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8 1606 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 1607 // CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1608 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 1609 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 1610 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1611 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1612 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1613 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1614 // CHECK9-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1615 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4 1616 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1617 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1618 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1619 // CHECK9: omp.precond.then: 1620 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1621 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1622 // CHECK9-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 1623 // CHECK9-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8 1624 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP7]] to i32 1625 // CHECK9-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 1626 // CHECK9-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP8]] to i32 1627 // CHECK9-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4 1628 // CHECK9-NEXT: store i32 [[CONV3]], ptr [[DOTOMP_UB]], align 4 1629 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1630 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1631 // CHECK9-NEXT: [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1632 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4 1633 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1634 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1635 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1636 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]] 1637 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1638 // CHECK9: cond.true: 1639 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1640 // CHECK9-NEXT: br label [[COND_END:%.*]] 1641 // CHECK9: cond.false: 1642 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1643 // CHECK9-NEXT: br label [[COND_END]] 1644 // CHECK9: cond.end: 1645 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ] 1646 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 1647 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1648 // CHECK9-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4 1649 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1650 // CHECK9: omp.inner.for.cond: 1651 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1652 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1653 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]] 1654 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1655 // CHECK9: omp.inner.for.body: 1656 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1657 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1 1658 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1659 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I4]], align 4 1660 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, ptr [[I4]], align 4 1661 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP19]] to i64 1662 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]] 1663 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 1664 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1665 // CHECK9: omp.body.continue: 1666 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1667 // CHECK9: omp.inner.for.inc: 1668 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1669 // CHECK9-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP20]], 1 1670 // CHECK9-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4 1671 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 1672 // CHECK9: omp.inner.for.end: 1673 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1674 // CHECK9: omp.loop.exit: 1675 // CHECK9-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1676 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4 1677 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]]) 1678 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 1679 // CHECK9: omp.precond.end: 1680 // CHECK9-NEXT: ret void 1681 // 1682 // 1683 // CHECK11-LABEL: define {{[^@]+}}@_Z15teams_local_argv 1684 // CHECK11-SAME: () #[[ATTR0:[0-9]+]] { 1685 // CHECK11-NEXT: entry: 1686 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 1687 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 4 1688 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 1689 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 1690 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4 1691 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4 1692 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4 1693 // CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4 1694 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 1695 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1696 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1697 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 1698 // CHECK11-NEXT: store i32 100, ptr [[N]], align 4 1699 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 1700 // CHECK11-NEXT: [[TMP1:%.*]] = call ptr @llvm.stacksave.p0() 1701 // CHECK11-NEXT: store ptr [[TMP1]], ptr [[SAVED_STACK]], align 4 1702 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 1703 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[__VLA_EXPR0]], align 4 1704 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[N]], align 4 1705 // CHECK11-NEXT: store i32 [[TMP2]], ptr [[N_CASTED]], align 4 1706 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_CASTED]], align 4 1707 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 1708 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 1709 // CHECK11-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 24, i1 false) 1710 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1711 // CHECK11-NEXT: store i32 [[TMP3]], ptr [[TMP6]], align 4 1712 // CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1713 // CHECK11-NEXT: store i32 [[TMP3]], ptr [[TMP7]], align 4 1714 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 1715 // CHECK11-NEXT: store ptr null, ptr [[TMP8]], align 4 1716 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 1717 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP9]], align 4 1718 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 1719 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP10]], align 4 1720 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 1721 // CHECK11-NEXT: store ptr null, ptr [[TMP11]], align 4 1722 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 1723 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP12]], align 4 1724 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 1725 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP13]], align 4 1726 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2 1727 // CHECK11-NEXT: store i64 [[TMP5]], ptr [[TMP14]], align 4 1728 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 1729 // CHECK11-NEXT: store ptr null, ptr [[TMP15]], align 4 1730 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1731 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1732 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 1733 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, ptr [[N]], align 4 1734 // CHECK11-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4 1735 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1736 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0 1737 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1738 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1739 // CHECK11-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1740 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1741 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], 1 1742 // CHECK11-NEXT: [[TMP22:%.*]] = zext i32 [[ADD]] to i64 1743 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 1744 // CHECK11-NEXT: store i32 3, ptr [[TMP23]], align 4 1745 // CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 1746 // CHECK11-NEXT: store i32 3, ptr [[TMP24]], align 4 1747 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 1748 // CHECK11-NEXT: store ptr [[TMP16]], ptr [[TMP25]], align 4 1749 // CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 1750 // CHECK11-NEXT: store ptr [[TMP17]], ptr [[TMP26]], align 4 1751 // CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 1752 // CHECK11-NEXT: store ptr [[TMP18]], ptr [[TMP27]], align 4 1753 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 1754 // CHECK11-NEXT: store ptr @.offload_maptypes, ptr [[TMP28]], align 4 1755 // CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 1756 // CHECK11-NEXT: store ptr null, ptr [[TMP29]], align 4 1757 // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 1758 // CHECK11-NEXT: store ptr null, ptr [[TMP30]], align 4 1759 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 1760 // CHECK11-NEXT: store i64 [[TMP22]], ptr [[TMP31]], align 8 1761 // CHECK11-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 1762 // CHECK11-NEXT: store i64 0, ptr [[TMP32]], align 8 1763 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 1764 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4 1765 // CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 1766 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP34]], align 4 1767 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 1768 // CHECK11-NEXT: store i32 0, ptr [[TMP35]], align 4 1769 // CHECK11-NEXT: [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, ptr [[KERNEL_ARGS]]) 1770 // CHECK11-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0 1771 // CHECK11-NEXT: br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1772 // CHECK11: omp_offload.failed: 1773 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i32 [[TMP3]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3:[0-9]+]] 1774 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] 1775 // CHECK11: omp_offload.cont: 1776 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i32 0 1777 // CHECK11-NEXT: [[TMP38:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 1778 // CHECK11-NEXT: [[TMP39:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4 1779 // CHECK11-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP39]]) 1780 // CHECK11-NEXT: ret i32 [[TMP38]] 1781 // 1782 // 1783 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73 1784 // CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 1785 // CHECK11-NEXT: entry: 1786 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 1787 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 1788 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 1789 // CHECK11-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 1790 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 1791 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 1792 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 1793 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 1794 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]]) 1795 // CHECK11-NEXT: ret void 1796 // 1797 // 1798 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.omp_outlined 1799 // 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]] { 1800 // CHECK11-NEXT: entry: 1801 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 1802 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 1803 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4 1804 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 1805 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 1806 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1807 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 1808 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1809 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1810 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 1811 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 1812 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 1813 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1814 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1815 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 1816 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 1817 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 1818 // CHECK11-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4 1819 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 1820 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 1821 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4 1822 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 1823 // CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4 1824 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 1825 // CHECK11-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 1826 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1827 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1828 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1829 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1830 // CHECK11-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1831 // CHECK11-NEXT: store i32 0, ptr [[I]], align 4 1832 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1833 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1834 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1835 // CHECK11: omp.precond.then: 1836 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 1837 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1838 // CHECK11-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4 1839 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1840 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1841 // CHECK11-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1842 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 1843 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1844 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1845 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1846 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 1847 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1848 // CHECK11: cond.true: 1849 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1850 // CHECK11-NEXT: br label [[COND_END:%.*]] 1851 // CHECK11: cond.false: 1852 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1853 // CHECK11-NEXT: br label [[COND_END]] 1854 // CHECK11: cond.end: 1855 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 1856 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 1857 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1858 // CHECK11-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 1859 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1860 // CHECK11: omp.inner.for.cond: 1861 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1862 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1863 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 1864 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1865 // CHECK11: omp.inner.for.body: 1866 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 1867 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 1868 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.omp_outlined.omp_outlined, i32 [[TMP16]], i32 [[TMP17]], ptr [[TMP0]], i32 [[TMP1]], ptr [[TMP2]]) 1869 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1870 // CHECK11: omp.inner.for.inc: 1871 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1872 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 1873 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] 1874 // CHECK11-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 1875 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 1876 // CHECK11: omp.inner.for.end: 1877 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1878 // CHECK11: omp.loop.exit: 1879 // CHECK11-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1880 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4 1881 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP21]]) 1882 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 1883 // CHECK11: omp.precond.end: 1884 // CHECK11-NEXT: ret void 1885 // 1886 // 1887 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.omp_outlined.omp_outlined 1888 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1889 // CHECK11-NEXT: entry: 1890 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 1891 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 1892 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 1893 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 1894 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4 1895 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 1896 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 1897 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1898 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 1899 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1900 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1901 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 1902 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1903 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1904 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1905 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1906 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 1907 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 1908 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 1909 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4 1910 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4 1911 // CHECK11-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4 1912 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 1913 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 1914 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4 1915 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 1916 // CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4 1917 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 1918 // CHECK11-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 1919 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1920 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1921 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1922 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1923 // CHECK11-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1924 // CHECK11-NEXT: store i32 0, ptr [[I]], align 4 1925 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1926 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1927 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1928 // CHECK11: omp.precond.then: 1929 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1930 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1931 // CHECK11-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 1932 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4 1933 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4 1934 // CHECK11-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_LB]], align 4 1935 // CHECK11-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_UB]], align 4 1936 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1937 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1938 // CHECK11-NEXT: [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1939 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4 1940 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1941 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1942 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1943 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]] 1944 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1945 // CHECK11: cond.true: 1946 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1947 // CHECK11-NEXT: br label [[COND_END:%.*]] 1948 // CHECK11: cond.false: 1949 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1950 // CHECK11-NEXT: br label [[COND_END]] 1951 // CHECK11: cond.end: 1952 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ] 1953 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 1954 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1955 // CHECK11-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4 1956 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1957 // CHECK11: omp.inner.for.cond: 1958 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1959 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1960 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]] 1961 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1962 // CHECK11: omp.inner.for.body: 1963 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1964 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1 1965 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1966 // CHECK11-NEXT: store i32 [[ADD]], ptr [[I3]], align 4 1967 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, ptr [[I3]], align 4 1968 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP19]] 1969 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 1970 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1971 // CHECK11: omp.body.continue: 1972 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1973 // CHECK11: omp.inner.for.inc: 1974 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1975 // CHECK11-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP20]], 1 1976 // CHECK11-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 1977 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 1978 // CHECK11: omp.inner.for.end: 1979 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1980 // CHECK11: omp.loop.exit: 1981 // CHECK11-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1982 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4 1983 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]]) 1984 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 1985 // CHECK11: omp.precond.end: 1986 // CHECK11-NEXT: ret void 1987 // 1988 // 1989 // CHECK17-LABEL: define {{[^@]+}}@_Z21teams_template_structv 1990 // CHECK17-SAME: () #[[ATTR0:[0-9]+]] { 1991 // CHECK17-NEXT: entry: 1992 // CHECK17-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 1993 // CHECK17-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]]) 1994 // CHECK17-NEXT: ret i32 [[CALL]] 1995 // 1996 // 1997 // CHECK17-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 1998 // CHECK17-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat { 1999 // CHECK17-NEXT: entry: 2000 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 2001 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 8 2002 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 8 2003 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 8 2004 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4 2005 // CHECK17-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 2006 // CHECK17-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 2007 // CHECK17-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 2008 // CHECK17-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 2009 // CHECK17-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2010 // CHECK17-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 8 2011 // CHECK17-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2012 // CHECK17-NEXT: store ptr [[A]], ptr [[TMP1]], align 8 2013 // CHECK17-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 2014 // CHECK17-NEXT: store ptr null, ptr [[TMP2]], align 8 2015 // CHECK17-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2016 // CHECK17-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2017 // CHECK17-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 2018 // CHECK17-NEXT: store i32 3, ptr [[TMP5]], align 4 2019 // CHECK17-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 2020 // CHECK17-NEXT: store i32 1, ptr [[TMP6]], align 4 2021 // CHECK17-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 2022 // CHECK17-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 8 2023 // CHECK17-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 2024 // CHECK17-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 8 2025 // CHECK17-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 2026 // CHECK17-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 8 2027 // CHECK17-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 2028 // CHECK17-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 8 2029 // CHECK17-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 2030 // CHECK17-NEXT: store ptr null, ptr [[TMP11]], align 8 2031 // CHECK17-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 2032 // CHECK17-NEXT: store ptr null, ptr [[TMP12]], align 8 2033 // CHECK17-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 2034 // CHECK17-NEXT: store i64 123, ptr [[TMP13]], align 8 2035 // CHECK17-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 2036 // CHECK17-NEXT: store i64 0, ptr [[TMP14]], align 8 2037 // CHECK17-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 2038 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4 2039 // CHECK17-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 2040 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4 2041 // CHECK17-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 2042 // CHECK17-NEXT: store i32 0, ptr [[TMP17]], align 4 2043 // CHECK17-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109.region_id, ptr [[KERNEL_ARGS]]) 2044 // CHECK17-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0 2045 // CHECK17-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2046 // CHECK17: omp_offload.failed: 2047 // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109(ptr [[THIS1]]) #[[ATTR2:[0-9]+]] 2048 // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]] 2049 // CHECK17: omp_offload.cont: 2050 // CHECK17-NEXT: [[A2:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 2051 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A2]], i64 0, i64 0 2052 // CHECK17-NEXT: [[TMP20:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 2053 // CHECK17-NEXT: ret i32 [[TMP20]] 2054 // 2055 // 2056 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109 2057 // CHECK17-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 2058 // CHECK17-NEXT: entry: 2059 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 2060 // CHECK17-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 2061 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 2062 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109.omp_outlined, ptr [[TMP0]]) 2063 // CHECK17-NEXT: ret void 2064 // 2065 // 2066 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109.omp_outlined 2067 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 2068 // CHECK17-NEXT: entry: 2069 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 2070 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 2071 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 2072 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2073 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4 2074 // CHECK17-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 2075 // CHECK17-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 2076 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2077 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2078 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4 2079 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 2080 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 2081 // CHECK17-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 2082 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 2083 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 2084 // CHECK17-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4 2085 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2086 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2087 // CHECK17-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2088 // CHECK17-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 2089 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2090 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2091 // CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 2092 // CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2093 // CHECK17: cond.true: 2094 // CHECK17-NEXT: br label [[COND_END:%.*]] 2095 // CHECK17: cond.false: 2096 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2097 // CHECK17-NEXT: br label [[COND_END]] 2098 // CHECK17: cond.end: 2099 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2100 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 2101 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2102 // CHECK17-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 2103 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2104 // CHECK17: omp.inner.for.cond: 2105 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2106 // CHECK17-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2107 // CHECK17-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 2108 // CHECK17-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2109 // CHECK17: omp.inner.for.body: 2110 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2111 // CHECK17-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 2112 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2113 // CHECK17-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 2114 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]]) 2115 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2116 // CHECK17: omp.inner.for.inc: 2117 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2118 // CHECK17-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 2119 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] 2120 // CHECK17-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 2121 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]] 2122 // CHECK17: omp.inner.for.end: 2123 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2124 // CHECK17: omp.loop.exit: 2125 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 2126 // CHECK17-NEXT: ret void 2127 // 2128 // 2129 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109.omp_outlined.omp_outlined 2130 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 2131 // CHECK17-NEXT: entry: 2132 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 2133 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 2134 // CHECK17-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 2135 // CHECK17-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 2136 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 2137 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2138 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4 2139 // CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2140 // CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2141 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2142 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2143 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4 2144 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 2145 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 2146 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8 2147 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8 2148 // CHECK17-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 2149 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 2150 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2151 // CHECK17-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 2152 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8 2153 // CHECK17-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32 2154 // CHECK17-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 2155 // CHECK17-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32 2156 // CHECK17-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4 2157 // CHECK17-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4 2158 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2159 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2160 // CHECK17-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2161 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 2162 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2163 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2164 // CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122 2165 // CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2166 // CHECK17: cond.true: 2167 // CHECK17-NEXT: br label [[COND_END:%.*]] 2168 // CHECK17: cond.false: 2169 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2170 // CHECK17-NEXT: br label [[COND_END]] 2171 // CHECK17: cond.end: 2172 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 2173 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2174 // CHECK17-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2175 // CHECK17-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 2176 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2177 // CHECK17: omp.inner.for.cond: 2178 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2179 // CHECK17-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2180 // CHECK17-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 2181 // CHECK17-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2182 // CHECK17: omp.inner.for.body: 2183 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2184 // CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 2185 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2186 // CHECK17-NEXT: store i32 [[ADD]], ptr [[I]], align 4 2187 // CHECK17-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 2188 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4 2189 // CHECK17-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 2190 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]] 2191 // CHECK17-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 2192 // CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2193 // CHECK17: omp.body.continue: 2194 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2195 // CHECK17: omp.inner.for.inc: 2196 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2197 // CHECK17-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1 2198 // CHECK17-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4 2199 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]] 2200 // CHECK17: omp.inner.for.end: 2201 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2202 // CHECK17: omp.loop.exit: 2203 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]]) 2204 // CHECK17-NEXT: ret void 2205 // 2206 // 2207 // CHECK19-LABEL: define {{[^@]+}}@_Z21teams_template_structv 2208 // CHECK19-SAME: () #[[ATTR0:[0-9]+]] { 2209 // CHECK19-NEXT: entry: 2210 // CHECK19-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 2211 // CHECK19-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]]) 2212 // CHECK19-NEXT: ret i32 [[CALL]] 2213 // 2214 // 2215 // CHECK19-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 2216 // CHECK19-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 2217 // CHECK19-NEXT: entry: 2218 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 2219 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x ptr], align 4 2220 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x ptr], align 4 2221 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x ptr], align 4 2222 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4 2223 // CHECK19-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 2224 // CHECK19-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 2225 // CHECK19-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 2226 // CHECK19-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 0 2227 // CHECK19-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2228 // CHECK19-NEXT: store ptr [[THIS1]], ptr [[TMP0]], align 4 2229 // CHECK19-NEXT: [[TMP1:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2230 // CHECK19-NEXT: store ptr [[A]], ptr [[TMP1]], align 4 2231 // CHECK19-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 2232 // CHECK19-NEXT: store ptr null, ptr [[TMP2]], align 4 2233 // CHECK19-NEXT: [[TMP3:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2234 // CHECK19-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2235 // CHECK19-NEXT: [[TMP5:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 2236 // CHECK19-NEXT: store i32 3, ptr [[TMP5]], align 4 2237 // CHECK19-NEXT: [[TMP6:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 2238 // CHECK19-NEXT: store i32 1, ptr [[TMP6]], align 4 2239 // CHECK19-NEXT: [[TMP7:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 2240 // CHECK19-NEXT: store ptr [[TMP3]], ptr [[TMP7]], align 4 2241 // CHECK19-NEXT: [[TMP8:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 2242 // CHECK19-NEXT: store ptr [[TMP4]], ptr [[TMP8]], align 4 2243 // CHECK19-NEXT: [[TMP9:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 2244 // CHECK19-NEXT: store ptr @.offload_sizes, ptr [[TMP9]], align 4 2245 // CHECK19-NEXT: [[TMP10:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 2246 // CHECK19-NEXT: store ptr @.offload_maptypes, ptr [[TMP10]], align 4 2247 // CHECK19-NEXT: [[TMP11:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 2248 // CHECK19-NEXT: store ptr null, ptr [[TMP11]], align 4 2249 // CHECK19-NEXT: [[TMP12:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 2250 // CHECK19-NEXT: store ptr null, ptr [[TMP12]], align 4 2251 // CHECK19-NEXT: [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 2252 // CHECK19-NEXT: store i64 123, ptr [[TMP13]], align 8 2253 // CHECK19-NEXT: [[TMP14:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 2254 // CHECK19-NEXT: store i64 0, ptr [[TMP14]], align 8 2255 // CHECK19-NEXT: [[TMP15:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 2256 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP15]], align 4 2257 // CHECK19-NEXT: [[TMP16:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 2258 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP16]], align 4 2259 // CHECK19-NEXT: [[TMP17:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 2260 // CHECK19-NEXT: store i32 0, ptr [[TMP17]], align 4 2261 // CHECK19-NEXT: [[TMP18:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109.region_id, ptr [[KERNEL_ARGS]]) 2262 // CHECK19-NEXT: [[TMP19:%.*]] = icmp ne i32 [[TMP18]], 0 2263 // CHECK19-NEXT: br i1 [[TMP19]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2264 // CHECK19: omp_offload.failed: 2265 // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109(ptr [[THIS1]]) #[[ATTR2:[0-9]+]] 2266 // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]] 2267 // CHECK19: omp_offload.cont: 2268 // CHECK19-NEXT: [[A2:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 2269 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A2]], i32 0, i32 0 2270 // CHECK19-NEXT: [[TMP20:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 2271 // CHECK19-NEXT: ret i32 [[TMP20]] 2272 // 2273 // 2274 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109 2275 // CHECK19-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 2276 // CHECK19-NEXT: entry: 2277 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 2278 // CHECK19-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 2279 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 2280 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109.omp_outlined, ptr [[TMP0]]) 2281 // CHECK19-NEXT: ret void 2282 // 2283 // 2284 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109.omp_outlined 2285 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 2286 // CHECK19-NEXT: entry: 2287 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 2288 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 2289 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 2290 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2291 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4 2292 // CHECK19-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 2293 // CHECK19-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 2294 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2295 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2296 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4 2297 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 2298 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 2299 // CHECK19-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 2300 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 2301 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 2302 // CHECK19-NEXT: store i32 122, ptr [[DOTOMP_COMB_UB]], align 4 2303 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2304 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2305 // CHECK19-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 2306 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 2307 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2308 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2309 // CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 2310 // CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2311 // CHECK19: cond.true: 2312 // CHECK19-NEXT: br label [[COND_END:%.*]] 2313 // CHECK19: cond.false: 2314 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2315 // CHECK19-NEXT: br label [[COND_END]] 2316 // CHECK19: cond.end: 2317 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2318 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 2319 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2320 // CHECK19-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 2321 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2322 // CHECK19: omp.inner.for.cond: 2323 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2324 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2325 // CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 2326 // CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2327 // CHECK19: omp.inner.for.body: 2328 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2329 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2330 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]]) 2331 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2332 // CHECK19: omp.inner.for.inc: 2333 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2334 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 2335 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] 2336 // CHECK19-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 2337 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]] 2338 // CHECK19: omp.inner.for.end: 2339 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2340 // CHECK19: omp.loop.exit: 2341 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 2342 // CHECK19-NEXT: ret void 2343 // 2344 // 2345 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109.omp_outlined.omp_outlined 2346 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 2347 // CHECK19-NEXT: entry: 2348 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 2349 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 2350 // CHECK19-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 2351 // CHECK19-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 2352 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 2353 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2354 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4 2355 // CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2356 // CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2357 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2358 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2359 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4 2360 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 2361 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 2362 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4 2363 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4 2364 // CHECK19-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 2365 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 2366 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2367 // CHECK19-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 2368 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4 2369 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4 2370 // CHECK19-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4 2371 // CHECK19-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4 2372 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2373 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2374 // CHECK19-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 2375 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 2376 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2377 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2378 // CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122 2379 // CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2380 // CHECK19: cond.true: 2381 // CHECK19-NEXT: br label [[COND_END:%.*]] 2382 // CHECK19: cond.false: 2383 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2384 // CHECK19-NEXT: br label [[COND_END]] 2385 // CHECK19: cond.end: 2386 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 2387 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2388 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2389 // CHECK19-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 2390 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2391 // CHECK19: omp.inner.for.cond: 2392 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2393 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2394 // CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 2395 // CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2396 // CHECK19: omp.inner.for.body: 2397 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2398 // CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 2399 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2400 // CHECK19-NEXT: store i32 [[ADD]], ptr [[I]], align 4 2401 // CHECK19-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 0 2402 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4 2403 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP11]] 2404 // CHECK19-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 2405 // CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2406 // CHECK19: omp.body.continue: 2407 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2408 // CHECK19: omp.inner.for.inc: 2409 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2410 // CHECK19-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1 2411 // CHECK19-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 2412 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]] 2413 // CHECK19: omp.inner.for.end: 2414 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2415 // CHECK19: omp.loop.exit: 2416 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]]) 2417 // CHECK19-NEXT: ret void 2418 // 2419 // 2420 // CHECK25-LABEL: define {{[^@]+}}@main 2421 // CHECK25-SAME: (i32 noundef signext [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 2422 // CHECK25-NEXT: entry: 2423 // CHECK25-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 2424 // CHECK25-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 2425 // CHECK25-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 8 2426 // CHECK25-NEXT: [[N:%.*]] = alloca i32, align 4 2427 // CHECK25-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 8 2428 // CHECK25-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 2429 // CHECK25-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 2430 // CHECK25-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8 2431 // CHECK25-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8 2432 // CHECK25-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8 2433 // CHECK25-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8 2434 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 2435 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2436 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2437 // CHECK25-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 2438 // CHECK25-NEXT: store i32 0, ptr [[RETVAL]], align 4 2439 // CHECK25-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 2440 // CHECK25-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8 2441 // CHECK25-NEXT: store i32 100, ptr [[N]], align 4 2442 // CHECK25-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 2443 // CHECK25-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 2444 // CHECK25-NEXT: [[TMP2:%.*]] = call ptr @llvm.stacksave.p0() 2445 // CHECK25-NEXT: store ptr [[TMP2]], ptr [[SAVED_STACK]], align 8 2446 // CHECK25-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 2447 // CHECK25-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8 2448 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, ptr [[N]], align 4 2449 // CHECK25-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 2450 // CHECK25-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8 2451 // CHECK25-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 2452 // CHECK25-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 24, i1 false) 2453 // CHECK25-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2454 // CHECK25-NEXT: store i64 [[TMP4]], ptr [[TMP6]], align 8 2455 // CHECK25-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2456 // CHECK25-NEXT: store i64 [[TMP4]], ptr [[TMP7]], align 8 2457 // CHECK25-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 2458 // CHECK25-NEXT: store ptr null, ptr [[TMP8]], align 8 2459 // CHECK25-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 2460 // CHECK25-NEXT: store i64 [[TMP1]], ptr [[TMP9]], align 8 2461 // CHECK25-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 2462 // CHECK25-NEXT: store i64 [[TMP1]], ptr [[TMP10]], align 8 2463 // CHECK25-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 2464 // CHECK25-NEXT: store ptr null, ptr [[TMP11]], align 8 2465 // CHECK25-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 2466 // CHECK25-NEXT: store ptr [[VLA]], ptr [[TMP12]], align 8 2467 // CHECK25-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 2468 // CHECK25-NEXT: store ptr [[VLA]], ptr [[TMP13]], align 8 2469 // CHECK25-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2 2470 // CHECK25-NEXT: store i64 [[TMP5]], ptr [[TMP14]], align 8 2471 // CHECK25-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 2472 // CHECK25-NEXT: store ptr null, ptr [[TMP15]], align 8 2473 // CHECK25-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2474 // CHECK25-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2475 // CHECK25-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 2476 // CHECK25-NEXT: [[TMP19:%.*]] = load i32, ptr [[N]], align 4 2477 // CHECK25-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4 2478 // CHECK25-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 2479 // CHECK25-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0 2480 // CHECK25-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2481 // CHECK25-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 2482 // CHECK25-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 2483 // CHECK25-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2484 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], 1 2485 // CHECK25-NEXT: [[TMP22:%.*]] = zext i32 [[ADD]] to i64 2486 // CHECK25-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 2487 // CHECK25-NEXT: store i32 3, ptr [[TMP23]], align 4 2488 // CHECK25-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 2489 // CHECK25-NEXT: store i32 3, ptr [[TMP24]], align 4 2490 // CHECK25-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 2491 // CHECK25-NEXT: store ptr [[TMP16]], ptr [[TMP25]], align 8 2492 // CHECK25-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 2493 // CHECK25-NEXT: store ptr [[TMP17]], ptr [[TMP26]], align 8 2494 // CHECK25-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 2495 // CHECK25-NEXT: store ptr [[TMP18]], ptr [[TMP27]], align 8 2496 // CHECK25-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 2497 // CHECK25-NEXT: store ptr @.offload_maptypes, ptr [[TMP28]], align 8 2498 // CHECK25-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 2499 // CHECK25-NEXT: store ptr null, ptr [[TMP29]], align 8 2500 // CHECK25-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 2501 // CHECK25-NEXT: store ptr null, ptr [[TMP30]], align 8 2502 // CHECK25-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 2503 // CHECK25-NEXT: store i64 [[TMP22]], ptr [[TMP31]], align 8 2504 // CHECK25-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 2505 // CHECK25-NEXT: store i64 0, ptr [[TMP32]], align 8 2506 // CHECK25-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 2507 // CHECK25-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4 2508 // CHECK25-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 2509 // CHECK25-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP34]], align 4 2510 // CHECK25-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 2511 // CHECK25-NEXT: store i32 0, ptr [[TMP35]], align 4 2512 // CHECK25-NEXT: [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, ptr [[KERNEL_ARGS]]) 2513 // CHECK25-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0 2514 // CHECK25-NEXT: br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2515 // CHECK25: omp_offload.failed: 2516 // CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i64 [[TMP4]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3:[0-9]+]] 2517 // CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]] 2518 // CHECK25: omp_offload.cont: 2519 // CHECK25-NEXT: [[TMP38:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 2520 // CHECK25-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10EEiT_(i32 noundef signext [[TMP38]]) 2521 // CHECK25-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 2522 // CHECK25-NEXT: [[TMP39:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 2523 // CHECK25-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP39]]) 2524 // CHECK25-NEXT: [[TMP40:%.*]] = load i32, ptr [[RETVAL]], align 4 2525 // CHECK25-NEXT: ret i32 [[TMP40]] 2526 // 2527 // 2528 // CHECK25-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162 2529 // CHECK25-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 2530 // CHECK25-NEXT: entry: 2531 // CHECK25-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 2532 // CHECK25-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 2533 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2534 // CHECK25-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 2535 // CHECK25-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 2536 // CHECK25-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2537 // CHECK25-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 2538 // CHECK25-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2539 // CHECK25-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]]) 2540 // CHECK25-NEXT: ret void 2541 // 2542 // 2543 // CHECK25-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.omp_outlined 2544 // CHECK25-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]] { 2545 // CHECK25-NEXT: entry: 2546 // CHECK25-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 2547 // CHECK25-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 2548 // CHECK25-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8 2549 // CHECK25-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 2550 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2551 // CHECK25-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2552 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 2553 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2554 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2555 // CHECK25-NEXT: [[I:%.*]] = alloca i32, align 4 2556 // CHECK25-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 2557 // CHECK25-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 2558 // CHECK25-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2559 // CHECK25-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2560 // CHECK25-NEXT: [[I3:%.*]] = alloca i32, align 4 2561 // CHECK25-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 2562 // CHECK25-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 2563 // CHECK25-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8 2564 // CHECK25-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 2565 // CHECK25-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2566 // CHECK25-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8 2567 // CHECK25-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 2568 // CHECK25-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2569 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 2570 // CHECK25-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 2571 // CHECK25-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 2572 // CHECK25-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 2573 // CHECK25-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2574 // CHECK25-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 2575 // CHECK25-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 2576 // CHECK25-NEXT: store i32 0, ptr [[I]], align 4 2577 // CHECK25-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 2578 // CHECK25-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 2579 // CHECK25-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2580 // CHECK25: omp.precond.then: 2581 // CHECK25-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 2582 // CHECK25-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2583 // CHECK25-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4 2584 // CHECK25-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2585 // CHECK25-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2586 // CHECK25-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2587 // CHECK25-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 2588 // CHECK25-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2589 // CHECK25-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2590 // CHECK25-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2591 // CHECK25-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 2592 // CHECK25-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2593 // CHECK25: cond.true: 2594 // CHECK25-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2595 // CHECK25-NEXT: br label [[COND_END:%.*]] 2596 // CHECK25: cond.false: 2597 // CHECK25-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2598 // CHECK25-NEXT: br label [[COND_END]] 2599 // CHECK25: cond.end: 2600 // CHECK25-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 2601 // CHECK25-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 2602 // CHECK25-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2603 // CHECK25-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 2604 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2605 // CHECK25: omp.inner.for.cond: 2606 // CHECK25-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2607 // CHECK25-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2608 // CHECK25-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 2609 // CHECK25-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2610 // CHECK25: omp.inner.for.body: 2611 // CHECK25-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2612 // CHECK25-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64 2613 // CHECK25-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2614 // CHECK25-NEXT: [[TMP19:%.*]] = zext i32 [[TMP18]] to i64 2615 // CHECK25-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.omp_outlined.omp_outlined, i64 [[TMP17]], i64 [[TMP19]], ptr [[TMP0]], i64 [[TMP1]], ptr [[TMP2]]) 2616 // CHECK25-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2617 // CHECK25: omp.inner.for.inc: 2618 // CHECK25-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2619 // CHECK25-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 2620 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]] 2621 // CHECK25-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 2622 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND]] 2623 // CHECK25: omp.inner.for.end: 2624 // CHECK25-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2625 // CHECK25: omp.loop.exit: 2626 // CHECK25-NEXT: [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2627 // CHECK25-NEXT: [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4 2628 // CHECK25-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP23]]) 2629 // CHECK25-NEXT: br label [[OMP_PRECOND_END]] 2630 // CHECK25: omp.precond.end: 2631 // CHECK25-NEXT: ret void 2632 // 2633 // 2634 // CHECK25-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.omp_outlined.omp_outlined 2635 // CHECK25-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 2636 // CHECK25-NEXT: entry: 2637 // CHECK25-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 2638 // CHECK25-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 2639 // CHECK25-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 2640 // CHECK25-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 2641 // CHECK25-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8 2642 // CHECK25-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 2643 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2644 // CHECK25-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2645 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 2646 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2647 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2648 // CHECK25-NEXT: [[I:%.*]] = alloca i32, align 4 2649 // CHECK25-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2650 // CHECK25-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2651 // CHECK25-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2652 // CHECK25-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2653 // CHECK25-NEXT: [[I4:%.*]] = alloca i32, align 4 2654 // CHECK25-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 2655 // CHECK25-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 2656 // CHECK25-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8 2657 // CHECK25-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8 2658 // CHECK25-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8 2659 // CHECK25-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 2660 // CHECK25-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2661 // CHECK25-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8 2662 // CHECK25-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 2663 // CHECK25-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2664 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 2665 // CHECK25-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 2666 // CHECK25-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 2667 // CHECK25-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 2668 // CHECK25-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2669 // CHECK25-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 2670 // CHECK25-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 2671 // CHECK25-NEXT: store i32 0, ptr [[I]], align 4 2672 // CHECK25-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 2673 // CHECK25-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 2674 // CHECK25-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2675 // CHECK25: omp.precond.then: 2676 // CHECK25-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2677 // CHECK25-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2678 // CHECK25-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 2679 // CHECK25-NEXT: [[TMP7:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8 2680 // CHECK25-NEXT: [[CONV:%.*]] = trunc i64 [[TMP7]] to i32 2681 // CHECK25-NEXT: [[TMP8:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 2682 // CHECK25-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP8]] to i32 2683 // CHECK25-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4 2684 // CHECK25-NEXT: store i32 [[CONV3]], ptr [[DOTOMP_UB]], align 4 2685 // CHECK25-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2686 // CHECK25-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2687 // CHECK25-NEXT: [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2688 // CHECK25-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4 2689 // CHECK25-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2690 // CHECK25-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2691 // CHECK25-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2692 // CHECK25-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]] 2693 // CHECK25-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2694 // CHECK25: cond.true: 2695 // CHECK25-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 2696 // CHECK25-NEXT: br label [[COND_END:%.*]] 2697 // CHECK25: cond.false: 2698 // CHECK25-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2699 // CHECK25-NEXT: br label [[COND_END]] 2700 // CHECK25: cond.end: 2701 // CHECK25-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ] 2702 // CHECK25-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2703 // CHECK25-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2704 // CHECK25-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4 2705 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2706 // CHECK25: omp.inner.for.cond: 2707 // CHECK25-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2708 // CHECK25-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2709 // CHECK25-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]] 2710 // CHECK25-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2711 // CHECK25: omp.inner.for.body: 2712 // CHECK25-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2713 // CHECK25-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1 2714 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2715 // CHECK25-NEXT: store i32 [[ADD]], ptr [[I4]], align 4 2716 // CHECK25-NEXT: [[TMP19:%.*]] = load i32, ptr [[I4]], align 4 2717 // CHECK25-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP19]] to i64 2718 // CHECK25-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]] 2719 // CHECK25-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 2720 // CHECK25-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2721 // CHECK25: omp.body.continue: 2722 // CHECK25-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2723 // CHECK25: omp.inner.for.inc: 2724 // CHECK25-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2725 // CHECK25-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP20]], 1 2726 // CHECK25-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4 2727 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND]] 2728 // CHECK25: omp.inner.for.end: 2729 // CHECK25-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2730 // CHECK25: omp.loop.exit: 2731 // CHECK25-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2732 // CHECK25-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4 2733 // CHECK25-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]]) 2734 // CHECK25-NEXT: br label [[OMP_PRECOND_END]] 2735 // CHECK25: omp.precond.end: 2736 // CHECK25-NEXT: ret void 2737 // 2738 // 2739 // CHECK25-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 2740 // CHECK25-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 2741 // CHECK25-NEXT: entry: 2742 // CHECK25-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 2743 // CHECK25-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 2744 // CHECK25-NEXT: [[TE:%.*]] = alloca i32, align 4 2745 // CHECK25-NEXT: [[TH:%.*]] = alloca i32, align 4 2746 // CHECK25-NEXT: [[TE_CASTED:%.*]] = alloca i64, align 8 2747 // CHECK25-NEXT: [[TH_CASTED:%.*]] = alloca i64, align 8 2748 // CHECK25-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8 2749 // CHECK25-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8 2750 // CHECK25-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8 2751 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 2752 // CHECK25-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 2753 // CHECK25-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 2754 // CHECK25-NEXT: store i32 0, ptr [[TE]], align 4 2755 // CHECK25-NEXT: store i32 128, ptr [[TH]], align 4 2756 // CHECK25-NEXT: [[TMP0:%.*]] = load i32, ptr [[TE]], align 4 2757 // CHECK25-NEXT: store i32 [[TMP0]], ptr [[TE_CASTED]], align 4 2758 // CHECK25-NEXT: [[TMP1:%.*]] = load i64, ptr [[TE_CASTED]], align 8 2759 // CHECK25-NEXT: [[TMP2:%.*]] = load i32, ptr [[TH]], align 4 2760 // CHECK25-NEXT: store i32 [[TMP2]], ptr [[TH_CASTED]], align 4 2761 // CHECK25-NEXT: [[TMP3:%.*]] = load i64, ptr [[TH_CASTED]], align 8 2762 // CHECK25-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2763 // CHECK25-NEXT: store i64 [[TMP1]], ptr [[TMP4]], align 8 2764 // CHECK25-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2765 // CHECK25-NEXT: store i64 [[TMP1]], ptr [[TMP5]], align 8 2766 // CHECK25-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 2767 // CHECK25-NEXT: store ptr null, ptr [[TMP6]], align 8 2768 // CHECK25-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 2769 // CHECK25-NEXT: store i64 [[TMP3]], ptr [[TMP7]], align 8 2770 // CHECK25-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 2771 // CHECK25-NEXT: store i64 [[TMP3]], ptr [[TMP8]], align 8 2772 // CHECK25-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 2773 // CHECK25-NEXT: store ptr null, ptr [[TMP9]], align 8 2774 // CHECK25-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 2775 // CHECK25-NEXT: store ptr [[A]], ptr [[TMP10]], align 8 2776 // CHECK25-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 2777 // CHECK25-NEXT: store ptr [[A]], ptr [[TMP11]], align 8 2778 // CHECK25-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 2779 // CHECK25-NEXT: store ptr null, ptr [[TMP12]], align 8 2780 // CHECK25-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2781 // CHECK25-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2782 // CHECK25-NEXT: [[TMP15:%.*]] = load i32, ptr [[TE]], align 4 2783 // CHECK25-NEXT: [[TMP16:%.*]] = load i32, ptr [[TH]], align 4 2784 // CHECK25-NEXT: [[TMP17:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP15]], 0 2785 // CHECK25-NEXT: [[TMP18:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP16]], 0 2786 // CHECK25-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 2787 // CHECK25-NEXT: store i32 3, ptr [[TMP19]], align 4 2788 // CHECK25-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 2789 // CHECK25-NEXT: store i32 3, ptr [[TMP20]], align 4 2790 // CHECK25-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 2791 // CHECK25-NEXT: store ptr [[TMP13]], ptr [[TMP21]], align 8 2792 // CHECK25-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 2793 // CHECK25-NEXT: store ptr [[TMP14]], ptr [[TMP22]], align 8 2794 // CHECK25-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 2795 // CHECK25-NEXT: store ptr @.offload_sizes.1, ptr [[TMP23]], align 8 2796 // CHECK25-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 2797 // CHECK25-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP24]], align 8 2798 // CHECK25-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 2799 // CHECK25-NEXT: store ptr null, ptr [[TMP25]], align 8 2800 // CHECK25-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 2801 // CHECK25-NEXT: store ptr null, ptr [[TMP26]], align 8 2802 // CHECK25-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 2803 // CHECK25-NEXT: store i64 10, ptr [[TMP27]], align 8 2804 // CHECK25-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 2805 // CHECK25-NEXT: store i64 0, ptr [[TMP28]], align 8 2806 // CHECK25-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 2807 // CHECK25-NEXT: store [3 x i32] [[TMP17]], ptr [[TMP29]], align 4 2808 // CHECK25-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 2809 // CHECK25-NEXT: store [3 x i32] [[TMP18]], ptr [[TMP30]], align 4 2810 // CHECK25-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 2811 // CHECK25-NEXT: store i32 0, ptr [[TMP31]], align 4 2812 // CHECK25-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 [[TMP15]], i32 [[TMP16]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, ptr [[KERNEL_ARGS]]) 2813 // CHECK25-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0 2814 // CHECK25-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2815 // CHECK25: omp_offload.failed: 2816 // CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i64 [[TMP1]], i64 [[TMP3]], ptr [[A]]) #[[ATTR3]] 2817 // CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]] 2818 // CHECK25: omp_offload.cont: 2819 // CHECK25-NEXT: ret i32 0 2820 // 2821 // 2822 // CHECK25-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151 2823 // CHECK25-SAME: (i64 noundef [[TE:%.*]], i64 noundef [[TH:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2824 // CHECK25-NEXT: entry: 2825 // CHECK25-NEXT: [[TE_ADDR:%.*]] = alloca i64, align 8 2826 // CHECK25-NEXT: [[TH_ADDR:%.*]] = alloca i64, align 8 2827 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2828 // CHECK25-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB3]]) 2829 // CHECK25-NEXT: store i64 [[TE]], ptr [[TE_ADDR]], align 8 2830 // CHECK25-NEXT: store i64 [[TH]], ptr [[TH_ADDR]], align 8 2831 // CHECK25-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2832 // CHECK25-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2833 // CHECK25-NEXT: [[TMP2:%.*]] = load i32, ptr [[TE_ADDR]], align 4 2834 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, ptr [[TH_ADDR]], align 4 2835 // CHECK25-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB3]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 2836 // CHECK25-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.omp_outlined, ptr [[TMP1]]) 2837 // CHECK25-NEXT: ret void 2838 // 2839 // 2840 // CHECK25-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.omp_outlined 2841 // CHECK25-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2842 // CHECK25-NEXT: entry: 2843 // CHECK25-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 2844 // CHECK25-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 2845 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2846 // CHECK25-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2847 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 2848 // CHECK25-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 2849 // CHECK25-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 2850 // CHECK25-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2851 // CHECK25-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2852 // CHECK25-NEXT: [[I:%.*]] = alloca i32, align 4 2853 // CHECK25-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 2854 // CHECK25-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 2855 // CHECK25-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2856 // CHECK25-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2857 // CHECK25-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 2858 // CHECK25-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4 2859 // CHECK25-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2860 // CHECK25-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2861 // CHECK25-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2862 // CHECK25-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 2863 // CHECK25-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2864 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2865 // CHECK25-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 2866 // CHECK25-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2867 // CHECK25: cond.true: 2868 // CHECK25-NEXT: br label [[COND_END:%.*]] 2869 // CHECK25: cond.false: 2870 // CHECK25-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2871 // CHECK25-NEXT: br label [[COND_END]] 2872 // CHECK25: cond.end: 2873 // CHECK25-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2874 // CHECK25-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 2875 // CHECK25-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2876 // CHECK25-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 2877 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2878 // CHECK25: omp.inner.for.cond: 2879 // CHECK25-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2880 // CHECK25-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2881 // CHECK25-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 2882 // CHECK25-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2883 // CHECK25: omp.inner.for.body: 2884 // CHECK25-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 2885 // CHECK25-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 2886 // CHECK25-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 2887 // CHECK25-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 2888 // CHECK25-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.omp_outlined.omp_outlined, i64 [[TMP9]], i64 [[TMP11]], ptr [[TMP0]]) 2889 // CHECK25-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2890 // CHECK25: omp.inner.for.inc: 2891 // CHECK25-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2892 // CHECK25-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 2893 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] 2894 // CHECK25-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 2895 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND]] 2896 // CHECK25: omp.inner.for.end: 2897 // CHECK25-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2898 // CHECK25: omp.loop.exit: 2899 // CHECK25-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 2900 // CHECK25-NEXT: ret void 2901 // 2902 // 2903 // CHECK25-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.omp_outlined.omp_outlined 2904 // CHECK25-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2905 // CHECK25-NEXT: entry: 2906 // CHECK25-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 2907 // CHECK25-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 2908 // CHECK25-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 2909 // CHECK25-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 2910 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 2911 // CHECK25-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2912 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 2913 // CHECK25-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2914 // CHECK25-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2915 // CHECK25-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2916 // CHECK25-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2917 // CHECK25-NEXT: [[I:%.*]] = alloca i32, align 4 2918 // CHECK25-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 2919 // CHECK25-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 2920 // CHECK25-NEXT: store i64 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 8 2921 // CHECK25-NEXT: store i64 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 8 2922 // CHECK25-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 2923 // CHECK25-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 2924 // CHECK25-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2925 // CHECK25-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 2926 // CHECK25-NEXT: [[TMP1:%.*]] = load i64, ptr [[DOTPREVIOUS_LB__ADDR]], align 8 2927 // CHECK25-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32 2928 // CHECK25-NEXT: [[TMP2:%.*]] = load i64, ptr [[DOTPREVIOUS_UB__ADDR]], align 8 2929 // CHECK25-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32 2930 // CHECK25-NEXT: store i32 [[CONV]], ptr [[DOTOMP_LB]], align 4 2931 // CHECK25-NEXT: store i32 [[CONV1]], ptr [[DOTOMP_UB]], align 4 2932 // CHECK25-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2933 // CHECK25-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2934 // CHECK25-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2935 // CHECK25-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 2936 // CHECK25-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2937 // CHECK25-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2938 // CHECK25-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9 2939 // CHECK25-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2940 // CHECK25: cond.true: 2941 // CHECK25-NEXT: br label [[COND_END:%.*]] 2942 // CHECK25: cond.false: 2943 // CHECK25-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2944 // CHECK25-NEXT: br label [[COND_END]] 2945 // CHECK25: cond.end: 2946 // CHECK25-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 2947 // CHECK25-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2948 // CHECK25-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2949 // CHECK25-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 2950 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2951 // CHECK25: omp.inner.for.cond: 2952 // CHECK25-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2953 // CHECK25-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2954 // CHECK25-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 2955 // CHECK25-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2956 // CHECK25: omp.inner.for.body: 2957 // CHECK25-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2958 // CHECK25-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 2959 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2960 // CHECK25-NEXT: store i32 [[ADD]], ptr [[I]], align 4 2961 // CHECK25-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4 2962 // CHECK25-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 2963 // CHECK25-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]] 2964 // CHECK25-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 2965 // CHECK25-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2966 // CHECK25: omp.body.continue: 2967 // CHECK25-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2968 // CHECK25: omp.inner.for.inc: 2969 // CHECK25-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2970 // CHECK25-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1 2971 // CHECK25-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4 2972 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND]] 2973 // CHECK25: omp.inner.for.end: 2974 // CHECK25-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2975 // CHECK25: omp.loop.exit: 2976 // CHECK25-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]]) 2977 // CHECK25-NEXT: ret void 2978 // 2979 // 2980 // CHECK27-LABEL: define {{[^@]+}}@main 2981 // CHECK27-SAME: (i32 noundef [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 2982 // CHECK27-NEXT: entry: 2983 // CHECK27-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 2984 // CHECK27-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 2985 // CHECK27-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 4 2986 // CHECK27-NEXT: [[N:%.*]] = alloca i32, align 4 2987 // CHECK27-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 4 2988 // CHECK27-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 2989 // CHECK27-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 2990 // CHECK27-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4 2991 // CHECK27-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4 2992 // CHECK27-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4 2993 // CHECK27-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4 2994 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 2995 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2996 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2997 // CHECK27-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 2998 // CHECK27-NEXT: store i32 0, ptr [[RETVAL]], align 4 2999 // CHECK27-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 3000 // CHECK27-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 4 3001 // CHECK27-NEXT: store i32 100, ptr [[N]], align 4 3002 // CHECK27-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 3003 // CHECK27-NEXT: [[TMP1:%.*]] = call ptr @llvm.stacksave.p0() 3004 // CHECK27-NEXT: store ptr [[TMP1]], ptr [[SAVED_STACK]], align 4 3005 // CHECK27-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 3006 // CHECK27-NEXT: store i32 [[TMP0]], ptr [[__VLA_EXPR0]], align 4 3007 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, ptr [[N]], align 4 3008 // CHECK27-NEXT: store i32 [[TMP2]], ptr [[N_CASTED]], align 4 3009 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_CASTED]], align 4 3010 // CHECK27-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 3011 // CHECK27-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 3012 // CHECK27-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 24, i1 false) 3013 // CHECK27-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3014 // CHECK27-NEXT: store i32 [[TMP3]], ptr [[TMP6]], align 4 3015 // CHECK27-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3016 // CHECK27-NEXT: store i32 [[TMP3]], ptr [[TMP7]], align 4 3017 // CHECK27-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 3018 // CHECK27-NEXT: store ptr null, ptr [[TMP8]], align 4 3019 // CHECK27-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 3020 // CHECK27-NEXT: store i32 [[TMP0]], ptr [[TMP9]], align 4 3021 // CHECK27-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 3022 // CHECK27-NEXT: store i32 [[TMP0]], ptr [[TMP10]], align 4 3023 // CHECK27-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 3024 // CHECK27-NEXT: store ptr null, ptr [[TMP11]], align 4 3025 // CHECK27-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 3026 // CHECK27-NEXT: store ptr [[VLA]], ptr [[TMP12]], align 4 3027 // CHECK27-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 3028 // CHECK27-NEXT: store ptr [[VLA]], ptr [[TMP13]], align 4 3029 // CHECK27-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2 3030 // CHECK27-NEXT: store i64 [[TMP5]], ptr [[TMP14]], align 4 3031 // CHECK27-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 3032 // CHECK27-NEXT: store ptr null, ptr [[TMP15]], align 4 3033 // CHECK27-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3034 // CHECK27-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3035 // CHECK27-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 3036 // CHECK27-NEXT: [[TMP19:%.*]] = load i32, ptr [[N]], align 4 3037 // CHECK27-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4 3038 // CHECK27-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3039 // CHECK27-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0 3040 // CHECK27-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3041 // CHECK27-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3042 // CHECK27-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 3043 // CHECK27-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3044 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], 1 3045 // CHECK27-NEXT: [[TMP22:%.*]] = zext i32 [[ADD]] to i64 3046 // CHECK27-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 3047 // CHECK27-NEXT: store i32 3, ptr [[TMP23]], align 4 3048 // CHECK27-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 3049 // CHECK27-NEXT: store i32 3, ptr [[TMP24]], align 4 3050 // CHECK27-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 3051 // CHECK27-NEXT: store ptr [[TMP16]], ptr [[TMP25]], align 4 3052 // CHECK27-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 3053 // CHECK27-NEXT: store ptr [[TMP17]], ptr [[TMP26]], align 4 3054 // CHECK27-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 3055 // CHECK27-NEXT: store ptr [[TMP18]], ptr [[TMP27]], align 4 3056 // CHECK27-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 3057 // CHECK27-NEXT: store ptr @.offload_maptypes, ptr [[TMP28]], align 4 3058 // CHECK27-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 3059 // CHECK27-NEXT: store ptr null, ptr [[TMP29]], align 4 3060 // CHECK27-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 3061 // CHECK27-NEXT: store ptr null, ptr [[TMP30]], align 4 3062 // CHECK27-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 3063 // CHECK27-NEXT: store i64 [[TMP22]], ptr [[TMP31]], align 8 3064 // CHECK27-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 3065 // CHECK27-NEXT: store i64 0, ptr [[TMP32]], align 8 3066 // CHECK27-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 3067 // CHECK27-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4 3068 // CHECK27-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 3069 // CHECK27-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP34]], align 4 3070 // CHECK27-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 3071 // CHECK27-NEXT: store i32 0, ptr [[TMP35]], align 4 3072 // CHECK27-NEXT: [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3:[0-9]+]], i64 -1, i32 0, i32 0, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, ptr [[KERNEL_ARGS]]) 3073 // CHECK27-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0 3074 // CHECK27-NEXT: br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 3075 // CHECK27: omp_offload.failed: 3076 // CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i32 [[TMP3]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3:[0-9]+]] 3077 // CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]] 3078 // CHECK27: omp_offload.cont: 3079 // CHECK27-NEXT: [[TMP38:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 3080 // CHECK27-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10EEiT_(i32 noundef [[TMP38]]) 3081 // CHECK27-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 3082 // CHECK27-NEXT: [[TMP39:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4 3083 // CHECK27-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP39]]) 3084 // CHECK27-NEXT: [[TMP40:%.*]] = load i32, ptr [[RETVAL]], align 4 3085 // CHECK27-NEXT: ret i32 [[TMP40]] 3086 // 3087 // 3088 // CHECK27-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162 3089 // CHECK27-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 3090 // CHECK27-NEXT: entry: 3091 // CHECK27-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 3092 // CHECK27-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 3093 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3094 // CHECK27-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 3095 // CHECK27-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 3096 // CHECK27-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3097 // CHECK27-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 3098 // CHECK27-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3099 // CHECK27-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]]) 3100 // CHECK27-NEXT: ret void 3101 // 3102 // 3103 // CHECK27-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.omp_outlined 3104 // CHECK27-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]] { 3105 // CHECK27-NEXT: entry: 3106 // CHECK27-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 3107 // CHECK27-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 3108 // CHECK27-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4 3109 // CHECK27-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 3110 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3111 // CHECK27-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3112 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 3113 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3114 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3115 // CHECK27-NEXT: [[I:%.*]] = alloca i32, align 4 3116 // CHECK27-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 3117 // CHECK27-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 3118 // CHECK27-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3119 // CHECK27-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3120 // CHECK27-NEXT: [[I3:%.*]] = alloca i32, align 4 3121 // CHECK27-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 3122 // CHECK27-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 3123 // CHECK27-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4 3124 // CHECK27-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 3125 // CHECK27-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3126 // CHECK27-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4 3127 // CHECK27-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 3128 // CHECK27-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3129 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 3130 // CHECK27-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 3131 // CHECK27-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3132 // CHECK27-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 3133 // CHECK27-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3134 // CHECK27-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3135 // CHECK27-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 3136 // CHECK27-NEXT: store i32 0, ptr [[I]], align 4 3137 // CHECK27-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3138 // CHECK27-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 3139 // CHECK27-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3140 // CHECK27: omp.precond.then: 3141 // CHECK27-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 3142 // CHECK27-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3143 // CHECK27-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_COMB_UB]], align 4 3144 // CHECK27-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 3145 // CHECK27-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 3146 // CHECK27-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3147 // CHECK27-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 3148 // CHECK27-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 3149 // CHECK27-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3150 // CHECK27-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3151 // CHECK27-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 3152 // CHECK27-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3153 // CHECK27: cond.true: 3154 // CHECK27-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3155 // CHECK27-NEXT: br label [[COND_END:%.*]] 3156 // CHECK27: cond.false: 3157 // CHECK27-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3158 // CHECK27-NEXT: br label [[COND_END]] 3159 // CHECK27: cond.end: 3160 // CHECK27-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 3161 // CHECK27-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 3162 // CHECK27-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 3163 // CHECK27-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 3164 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3165 // CHECK27: omp.inner.for.cond: 3166 // CHECK27-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3167 // CHECK27-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3168 // CHECK27-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 3169 // CHECK27-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3170 // CHECK27: omp.inner.for.body: 3171 // CHECK27-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 3172 // CHECK27-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3173 // CHECK27-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.omp_outlined.omp_outlined, i32 [[TMP16]], i32 [[TMP17]], ptr [[TMP0]], i32 [[TMP1]], ptr [[TMP2]]) 3174 // CHECK27-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3175 // CHECK27: omp.inner.for.inc: 3176 // CHECK27-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3177 // CHECK27-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 3178 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] 3179 // CHECK27-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 3180 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND]] 3181 // CHECK27: omp.inner.for.end: 3182 // CHECK27-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3183 // CHECK27: omp.loop.exit: 3184 // CHECK27-NEXT: [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3185 // CHECK27-NEXT: [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4 3186 // CHECK27-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP21]]) 3187 // CHECK27-NEXT: br label [[OMP_PRECOND_END]] 3188 // CHECK27: omp.precond.end: 3189 // CHECK27-NEXT: ret void 3190 // 3191 // 3192 // CHECK27-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.omp_outlined.omp_outlined 3193 // CHECK27-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 3194 // CHECK27-NEXT: entry: 3195 // CHECK27-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 3196 // CHECK27-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 3197 // CHECK27-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 3198 // CHECK27-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 3199 // CHECK27-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4 3200 // CHECK27-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 3201 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3202 // CHECK27-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3203 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 3204 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3205 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3206 // CHECK27-NEXT: [[I:%.*]] = alloca i32, align 4 3207 // CHECK27-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3208 // CHECK27-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3209 // CHECK27-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3210 // CHECK27-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3211 // CHECK27-NEXT: [[I3:%.*]] = alloca i32, align 4 3212 // CHECK27-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 3213 // CHECK27-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 3214 // CHECK27-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4 3215 // CHECK27-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4 3216 // CHECK27-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4 3217 // CHECK27-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 3218 // CHECK27-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3219 // CHECK27-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4 3220 // CHECK27-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 3221 // CHECK27-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3222 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 3223 // CHECK27-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 3224 // CHECK27-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3225 // CHECK27-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 3226 // CHECK27-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3227 // CHECK27-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3228 // CHECK27-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 3229 // CHECK27-NEXT: store i32 0, ptr [[I]], align 4 3230 // CHECK27-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3231 // CHECK27-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 3232 // CHECK27-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3233 // CHECK27: omp.precond.then: 3234 // CHECK27-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 3235 // CHECK27-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3236 // CHECK27-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 3237 // CHECK27-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4 3238 // CHECK27-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4 3239 // CHECK27-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_LB]], align 4 3240 // CHECK27-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_UB]], align 4 3241 // CHECK27-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 3242 // CHECK27-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 3243 // CHECK27-NEXT: [[TMP9:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3244 // CHECK27-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP9]], align 4 3245 // CHECK27-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 3246 // CHECK27-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3247 // CHECK27-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3248 // CHECK27-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]] 3249 // CHECK27-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3250 // CHECK27: cond.true: 3251 // CHECK27-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3252 // CHECK27-NEXT: br label [[COND_END:%.*]] 3253 // CHECK27: cond.false: 3254 // CHECK27-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3255 // CHECK27-NEXT: br label [[COND_END]] 3256 // CHECK27: cond.end: 3257 // CHECK27-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ] 3258 // CHECK27-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 3259 // CHECK27-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 3260 // CHECK27-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4 3261 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3262 // CHECK27: omp.inner.for.cond: 3263 // CHECK27-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3264 // CHECK27-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3265 // CHECK27-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]] 3266 // CHECK27-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3267 // CHECK27: omp.inner.for.body: 3268 // CHECK27-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3269 // CHECK27-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1 3270 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3271 // CHECK27-NEXT: store i32 [[ADD]], ptr [[I3]], align 4 3272 // CHECK27-NEXT: [[TMP19:%.*]] = load i32, ptr [[I3]], align 4 3273 // CHECK27-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP19]] 3274 // CHECK27-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 3275 // CHECK27-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3276 // CHECK27: omp.body.continue: 3277 // CHECK27-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3278 // CHECK27: omp.inner.for.inc: 3279 // CHECK27-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3280 // CHECK27-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP20]], 1 3281 // CHECK27-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 3282 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND]] 3283 // CHECK27: omp.inner.for.end: 3284 // CHECK27-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3285 // CHECK27: omp.loop.exit: 3286 // CHECK27-NEXT: [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3287 // CHECK27-NEXT: [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4 3288 // CHECK27-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP22]]) 3289 // CHECK27-NEXT: br label [[OMP_PRECOND_END]] 3290 // CHECK27: omp.precond.end: 3291 // CHECK27-NEXT: ret void 3292 // 3293 // 3294 // CHECK27-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 3295 // CHECK27-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 3296 // CHECK27-NEXT: entry: 3297 // CHECK27-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 3298 // CHECK27-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 3299 // CHECK27-NEXT: [[TE:%.*]] = alloca i32, align 4 3300 // CHECK27-NEXT: [[TH:%.*]] = alloca i32, align 4 3301 // CHECK27-NEXT: [[TE_CASTED:%.*]] = alloca i32, align 4 3302 // CHECK27-NEXT: [[TH_CASTED:%.*]] = alloca i32, align 4 3303 // CHECK27-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4 3304 // CHECK27-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4 3305 // CHECK27-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4 3306 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 3307 // CHECK27-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 3308 // CHECK27-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 3309 // CHECK27-NEXT: store i32 0, ptr [[TE]], align 4 3310 // CHECK27-NEXT: store i32 128, ptr [[TH]], align 4 3311 // CHECK27-NEXT: [[TMP0:%.*]] = load i32, ptr [[TE]], align 4 3312 // CHECK27-NEXT: store i32 [[TMP0]], ptr [[TE_CASTED]], align 4 3313 // CHECK27-NEXT: [[TMP1:%.*]] = load i32, ptr [[TE_CASTED]], align 4 3314 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, ptr [[TH]], align 4 3315 // CHECK27-NEXT: store i32 [[TMP2]], ptr [[TH_CASTED]], align 4 3316 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, ptr [[TH_CASTED]], align 4 3317 // CHECK27-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3318 // CHECK27-NEXT: store i32 [[TMP1]], ptr [[TMP4]], align 4 3319 // CHECK27-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3320 // CHECK27-NEXT: store i32 [[TMP1]], ptr [[TMP5]], align 4 3321 // CHECK27-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 3322 // CHECK27-NEXT: store ptr null, ptr [[TMP6]], align 4 3323 // CHECK27-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 3324 // CHECK27-NEXT: store i32 [[TMP3]], ptr [[TMP7]], align 4 3325 // CHECK27-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 3326 // CHECK27-NEXT: store i32 [[TMP3]], ptr [[TMP8]], align 4 3327 // CHECK27-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 3328 // CHECK27-NEXT: store ptr null, ptr [[TMP9]], align 4 3329 // CHECK27-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 3330 // CHECK27-NEXT: store ptr [[A]], ptr [[TMP10]], align 4 3331 // CHECK27-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 3332 // CHECK27-NEXT: store ptr [[A]], ptr [[TMP11]], align 4 3333 // CHECK27-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 3334 // CHECK27-NEXT: store ptr null, ptr [[TMP12]], align 4 3335 // CHECK27-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3336 // CHECK27-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3337 // CHECK27-NEXT: [[TMP15:%.*]] = load i32, ptr [[TE]], align 4 3338 // CHECK27-NEXT: [[TMP16:%.*]] = load i32, ptr [[TH]], align 4 3339 // CHECK27-NEXT: [[TMP17:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP15]], 0 3340 // CHECK27-NEXT: [[TMP18:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP16]], 0 3341 // CHECK27-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 3342 // CHECK27-NEXT: store i32 3, ptr [[TMP19]], align 4 3343 // CHECK27-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 3344 // CHECK27-NEXT: store i32 3, ptr [[TMP20]], align 4 3345 // CHECK27-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 3346 // CHECK27-NEXT: store ptr [[TMP13]], ptr [[TMP21]], align 4 3347 // CHECK27-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 3348 // CHECK27-NEXT: store ptr [[TMP14]], ptr [[TMP22]], align 4 3349 // CHECK27-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 3350 // CHECK27-NEXT: store ptr @.offload_sizes.1, ptr [[TMP23]], align 4 3351 // CHECK27-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 3352 // CHECK27-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP24]], align 4 3353 // CHECK27-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 3354 // CHECK27-NEXT: store ptr null, ptr [[TMP25]], align 4 3355 // CHECK27-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 3356 // CHECK27-NEXT: store ptr null, ptr [[TMP26]], align 4 3357 // CHECK27-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 3358 // CHECK27-NEXT: store i64 10, ptr [[TMP27]], align 8 3359 // CHECK27-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 3360 // CHECK27-NEXT: store i64 0, ptr [[TMP28]], align 8 3361 // CHECK27-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 3362 // CHECK27-NEXT: store [3 x i32] [[TMP17]], ptr [[TMP29]], align 4 3363 // CHECK27-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 3364 // CHECK27-NEXT: store [3 x i32] [[TMP18]], ptr [[TMP30]], align 4 3365 // CHECK27-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 3366 // CHECK27-NEXT: store i32 0, ptr [[TMP31]], align 4 3367 // CHECK27-NEXT: [[TMP32:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB3]], i64 -1, i32 [[TMP15]], i32 [[TMP16]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, ptr [[KERNEL_ARGS]]) 3368 // CHECK27-NEXT: [[TMP33:%.*]] = icmp ne i32 [[TMP32]], 0 3369 // CHECK27-NEXT: br i1 [[TMP33]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 3370 // CHECK27: omp_offload.failed: 3371 // CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i32 [[TMP1]], i32 [[TMP3]], ptr [[A]]) #[[ATTR3]] 3372 // CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]] 3373 // CHECK27: omp_offload.cont: 3374 // CHECK27-NEXT: ret i32 0 3375 // 3376 // 3377 // CHECK27-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151 3378 // CHECK27-SAME: (i32 noundef [[TE:%.*]], i32 noundef [[TH:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 3379 // CHECK27-NEXT: entry: 3380 // CHECK27-NEXT: [[TE_ADDR:%.*]] = alloca i32, align 4 3381 // CHECK27-NEXT: [[TH_ADDR:%.*]] = alloca i32, align 4 3382 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3383 // CHECK27-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB3]]) 3384 // CHECK27-NEXT: store i32 [[TE]], ptr [[TE_ADDR]], align 4 3385 // CHECK27-NEXT: store i32 [[TH]], ptr [[TH_ADDR]], align 4 3386 // CHECK27-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3387 // CHECK27-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3388 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, ptr [[TE_ADDR]], align 4 3389 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, ptr [[TH_ADDR]], align 4 3390 // CHECK27-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB3]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 3391 // CHECK27-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB3]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.omp_outlined, ptr [[TMP1]]) 3392 // CHECK27-NEXT: ret void 3393 // 3394 // 3395 // CHECK27-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.omp_outlined 3396 // CHECK27-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 3397 // CHECK27-NEXT: entry: 3398 // CHECK27-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 3399 // CHECK27-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 3400 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3401 // CHECK27-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3402 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 3403 // CHECK27-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 3404 // CHECK27-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 3405 // CHECK27-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3406 // CHECK27-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3407 // CHECK27-NEXT: [[I:%.*]] = alloca i32, align 4 3408 // CHECK27-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 3409 // CHECK27-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 3410 // CHECK27-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3411 // CHECK27-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3412 // CHECK27-NEXT: store i32 0, ptr [[DOTOMP_COMB_LB]], align 4 3413 // CHECK27-NEXT: store i32 9, ptr [[DOTOMP_COMB_UB]], align 4 3414 // CHECK27-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 3415 // CHECK27-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 3416 // CHECK27-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3417 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 3418 // CHECK27-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_COMB_LB]], ptr [[DOTOMP_COMB_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 3419 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3420 // CHECK27-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 3421 // CHECK27-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3422 // CHECK27: cond.true: 3423 // CHECK27-NEXT: br label [[COND_END:%.*]] 3424 // CHECK27: cond.false: 3425 // CHECK27-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3426 // CHECK27-NEXT: br label [[COND_END]] 3427 // CHECK27: cond.end: 3428 // CHECK27-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 3429 // CHECK27-NEXT: store i32 [[COND]], ptr [[DOTOMP_COMB_UB]], align 4 3430 // CHECK27-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 3431 // CHECK27-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 3432 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3433 // CHECK27: omp.inner.for.cond: 3434 // CHECK27-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3435 // CHECK27-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3436 // CHECK27-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 3437 // CHECK27-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3438 // CHECK27: omp.inner.for.body: 3439 // CHECK27-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_COMB_LB]], align 4 3440 // CHECK27-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_COMB_UB]], align 4 3441 // CHECK27-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.omp_outlined.omp_outlined, i32 [[TMP8]], i32 [[TMP9]], ptr [[TMP0]]) 3442 // CHECK27-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3443 // CHECK27: omp.inner.for.inc: 3444 // CHECK27-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3445 // CHECK27-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_STRIDE]], align 4 3446 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] 3447 // CHECK27-NEXT: store i32 [[ADD]], ptr [[DOTOMP_IV]], align 4 3448 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND]] 3449 // CHECK27: omp.inner.for.end: 3450 // CHECK27-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3451 // CHECK27: omp.loop.exit: 3452 // CHECK27-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 3453 // CHECK27-NEXT: ret void 3454 // 3455 // 3456 // CHECK27-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.omp_outlined.omp_outlined 3457 // CHECK27-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 3458 // CHECK27-NEXT: entry: 3459 // CHECK27-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 3460 // CHECK27-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 3461 // CHECK27-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 3462 // CHECK27-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 3463 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3464 // CHECK27-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3465 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 3466 // CHECK27-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3467 // CHECK27-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3468 // CHECK27-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3469 // CHECK27-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3470 // CHECK27-NEXT: [[I:%.*]] = alloca i32, align 4 3471 // CHECK27-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 3472 // CHECK27-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 3473 // CHECK27-NEXT: store i32 [[DOTPREVIOUS_LB_]], ptr [[DOTPREVIOUS_LB__ADDR]], align 4 3474 // CHECK27-NEXT: store i32 [[DOTPREVIOUS_UB_]], ptr [[DOTPREVIOUS_UB__ADDR]], align 4 3475 // CHECK27-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3476 // CHECK27-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3477 // CHECK27-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 3478 // CHECK27-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 3479 // CHECK27-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTPREVIOUS_LB__ADDR]], align 4 3480 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTPREVIOUS_UB__ADDR]], align 4 3481 // CHECK27-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_LB]], align 4 3482 // CHECK27-NEXT: store i32 [[TMP2]], ptr [[DOTOMP_UB]], align 4 3483 // CHECK27-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 3484 // CHECK27-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 3485 // CHECK27-NEXT: [[TMP3:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3486 // CHECK27-NEXT: [[TMP4:%.*]] = load i32, ptr [[TMP3]], align 4 3487 // CHECK27-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB2]], i32 [[TMP4]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 3488 // CHECK27-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3489 // CHECK27-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9 3490 // CHECK27-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3491 // CHECK27: cond.true: 3492 // CHECK27-NEXT: br label [[COND_END:%.*]] 3493 // CHECK27: cond.false: 3494 // CHECK27-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3495 // CHECK27-NEXT: br label [[COND_END]] 3496 // CHECK27: cond.end: 3497 // CHECK27-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 3498 // CHECK27-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 3499 // CHECK27-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 3500 // CHECK27-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 3501 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3502 // CHECK27: omp.inner.for.cond: 3503 // CHECK27-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3504 // CHECK27-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3505 // CHECK27-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 3506 // CHECK27-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3507 // CHECK27: omp.inner.for.body: 3508 // CHECK27-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3509 // CHECK27-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 3510 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3511 // CHECK27-NEXT: store i32 [[ADD]], ptr [[I]], align 4 3512 // CHECK27-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4 3513 // CHECK27-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP11]] 3514 // CHECK27-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 3515 // CHECK27-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3516 // CHECK27: omp.body.continue: 3517 // CHECK27-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3518 // CHECK27: omp.inner.for.inc: 3519 // CHECK27-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3520 // CHECK27-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1 3521 // CHECK27-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4 3522 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND]] 3523 // CHECK27: omp.inner.for.end: 3524 // CHECK27-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3525 // CHECK27: omp.loop.exit: 3526 // CHECK27-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB2]], i32 [[TMP4]]) 3527 // CHECK27-NEXT: ret void 3528 // 3529