1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _ 2 // expected-no-diagnostics 3 #ifndef HEADER 4 #define HEADER 5 // Test host codegen. 6 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1 7 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 8 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1 9 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3 10 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 11 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK3 12 13 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 14 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 15 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 16 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 17 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 18 // RUN: %clang_cc1 -no-opaque-pointers -DCK1 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 19 #ifdef CK1 20 21 int a[100]; 22 23 int teams_argument_global(int n){ 24 int te = n / 128; 25 int th = 128; 26 // discard n_addr 27 28 #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 -no-opaque-pointers -DCK2 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK9 55 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 56 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK9 57 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK11 58 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 59 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK11 60 61 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 62 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 63 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 64 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 65 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 66 // RUN: %clang_cc1 -no-opaque-pointers -DCK2 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 67 #ifdef CK2 68 69 int teams_local_arg(void) { 70 int n = 100; 71 int a[n]; 72 73 #pragma omp target 74 #pragma omp teams distribute 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 -no-opaque-pointers -DCK3 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK17 88 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 89 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK17 90 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK19 91 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 92 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK19 93 94 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 95 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 96 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 97 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 98 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 99 // RUN: %clang_cc1 -no-opaque-pointers -DCK3 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 100 #ifdef CK3 101 102 103 template <typename T, int X, long long Y> 104 struct SS{ 105 T a[X]; 106 float b; 107 int foo(void) { 108 109 #pragma omp target 110 #pragma omp teams distribute 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 -no-opaque-pointers -DCK4 -verify -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK25 131 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 132 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK25 133 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -verify -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK27 134 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 135 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK27 136 137 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -verify -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 138 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 139 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 140 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -verify -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 141 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 142 // RUN: %clang_cc1 -no-opaque-pointers -DCK4 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}" 143 144 #ifdef CK4 145 146 template <typename T, int n> 147 int tmain(T argc) { 148 T a[n]; 149 int te = n/128; 150 int th = 128; 151 #pragma omp target 152 #pragma omp teams distribute 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 i8*], align 8 188 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 8 189 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 8 190 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 191 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 192 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_3:%.*]] = alloca i32, align 4 193 // CHECK1-NEXT: [[N_CASTED6:%.*]] = alloca i64, align 8 194 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS8:%.*]] = alloca [2 x i8*], align 8 195 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS9:%.*]] = alloca [2 x i8*], align 8 196 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS10:%.*]] = alloca [2 x i8*], align 8 197 // CHECK1-NEXT: [[_TMP11:%.*]] = alloca i32, align 4 198 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_12:%.*]] = alloca i32, align 4 199 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_13:%.*]] = alloca i32, align 4 200 // CHECK1-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 201 // CHECK1-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 202 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP0]], 128 203 // CHECK1-NEXT: store i32 [[DIV]], i32* [[TE]], align 4 204 // CHECK1-NEXT: store i32 128, i32* [[TH]], align 4 205 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, i32* [[TE]], align 4 206 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[TE_CASTED]] to i32* 207 // CHECK1-NEXT: store i32 [[TMP1]], i32* [[CONV]], align 4 208 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, i64* [[TE_CASTED]], align 8 209 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[TH]], align 4 210 // CHECK1-NEXT: [[CONV1:%.*]] = bitcast i64* [[TH_CASTED]] to i32* 211 // CHECK1-NEXT: store i32 [[TMP3]], i32* [[CONV1]], align 4 212 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, i64* [[TH_CASTED]], align 8 213 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[N_ADDR]], align 4 214 // CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_CASTED]] to i32* 215 // CHECK1-NEXT: store i32 [[TMP5]], i32* [[CONV2]], align 4 216 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, i64* [[N_CASTED]], align 8 217 // CHECK1-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 218 // CHECK1-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64* 219 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP8]], align 8 220 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 221 // CHECK1-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64* 222 // CHECK1-NEXT: store i64 [[TMP2]], i64* [[TMP10]], align 8 223 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 224 // CHECK1-NEXT: store i8* null, i8** [[TMP11]], align 8 225 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 226 // CHECK1-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* 227 // CHECK1-NEXT: store i64 [[TMP4]], i64* [[TMP13]], align 8 228 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 229 // CHECK1-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* 230 // CHECK1-NEXT: store i64 [[TMP4]], i64* [[TMP15]], align 8 231 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 232 // CHECK1-NEXT: store i8* null, i8** [[TMP16]], align 8 233 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 234 // CHECK1-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i64* 235 // CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP18]], align 8 236 // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 237 // CHECK1-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i64* 238 // CHECK1-NEXT: store i64 [[TMP6]], i64* [[TMP20]], align 8 239 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 240 // CHECK1-NEXT: store i8* null, i8** [[TMP21]], align 8 241 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 242 // CHECK1-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to [100 x i32]** 243 // CHECK1-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP23]], align 8 244 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 245 // CHECK1-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [100 x i32]** 246 // CHECK1-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP25]], align 8 247 // CHECK1-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 248 // CHECK1-NEXT: store i8* null, i8** [[TMP26]], align 8 249 // CHECK1-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 250 // CHECK1-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 251 // CHECK1-NEXT: [[TMP29:%.*]] = load i32, i32* [[TE]], align 4 252 // CHECK1-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_ADDR]], align 4 253 // CHECK1-NEXT: store i32 [[TMP30]], i32* [[DOTCAPTURE_EXPR_]], align 4 254 // CHECK1-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 255 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP31]], 0 256 // CHECK1-NEXT: [[DIV4:%.*]] = sdiv i32 [[SUB]], 1 257 // CHECK1-NEXT: [[SUB5:%.*]] = sub nsw i32 [[DIV4]], 1 258 // CHECK1-NEXT: store i32 [[SUB5]], i32* [[DOTCAPTURE_EXPR_3]], align 4 259 // CHECK1-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_3]], align 4 260 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP32]], 1 261 // CHECK1-NEXT: [[TMP33:%.*]] = zext i32 [[ADD]] to i64 262 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP33]]) 263 // CHECK1-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 [[TMP29]], i32 0) 264 // CHECK1-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0 265 // CHECK1-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 266 // CHECK1: omp_offload.failed: 267 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28(i64 [[TMP2]], i64 [[TMP4]], i64 [[TMP6]], [100 x i32]* @a) #[[ATTR2:[0-9]+]] 268 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] 269 // CHECK1: omp_offload.cont: 270 // CHECK1-NEXT: [[TMP36:%.*]] = load i32, i32* [[N_ADDR]], align 4 271 // CHECK1-NEXT: [[CONV7:%.*]] = bitcast i64* [[N_CASTED6]] to i32* 272 // CHECK1-NEXT: store i32 [[TMP36]], i32* [[CONV7]], align 4 273 // CHECK1-NEXT: [[TMP37:%.*]] = load i64, i64* [[N_CASTED6]], align 8 274 // CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0 275 // CHECK1-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i64* 276 // CHECK1-NEXT: store i64 [[TMP37]], i64* [[TMP39]], align 8 277 // CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS9]], i32 0, i32 0 278 // CHECK1-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i64* 279 // CHECK1-NEXT: store i64 [[TMP37]], i64* [[TMP41]], align 8 280 // CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS10]], i64 0, i64 0 281 // CHECK1-NEXT: store i8* null, i8** [[TMP42]], align 8 282 // CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 1 283 // CHECK1-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to [100 x i32]** 284 // CHECK1-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP44]], align 8 285 // CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS9]], i32 0, i32 1 286 // CHECK1-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to [100 x i32]** 287 // CHECK1-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP46]], align 8 288 // CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS10]], i64 0, i64 1 289 // CHECK1-NEXT: store i8* null, i8** [[TMP47]], align 8 290 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS8]], i32 0, i32 0 291 // CHECK1-NEXT: [[TMP49:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS9]], i32 0, i32 0 292 // CHECK1-NEXT: [[TMP50:%.*]] = load i32, i32* [[N_ADDR]], align 4 293 // CHECK1-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_12]], align 4 294 // CHECK1-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_12]], align 4 295 // CHECK1-NEXT: [[SUB14:%.*]] = sub nsw i32 [[TMP51]], 0 296 // CHECK1-NEXT: [[DIV15:%.*]] = sdiv i32 [[SUB14]], 1 297 // CHECK1-NEXT: [[SUB16:%.*]] = sub nsw i32 [[DIV15]], 1 298 // CHECK1-NEXT: store i32 [[SUB16]], i32* [[DOTCAPTURE_EXPR_13]], align 4 299 // CHECK1-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_13]], align 4 300 // CHECK1-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP52]], 1 301 // CHECK1-NEXT: [[TMP53:%.*]] = zext i32 [[ADD17]] to i64 302 // CHECK1-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP53]]) 303 // CHECK1-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35.region_id, i32 2, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) 304 // CHECK1-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0 305 // CHECK1-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED18:%.*]], label [[OMP_OFFLOAD_CONT19:%.*]] 306 // CHECK1: omp_offload.failed18: 307 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35(i64 [[TMP37]], [100 x i32]* @a) #[[ATTR2]] 308 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT19]] 309 // CHECK1: omp_offload.cont19: 310 // CHECK1-NEXT: [[TMP56:%.*]] = load i32, i32* getelementptr inbounds ([100 x i32], [100 x i32]* @a, i64 0, i64 0), align 4 311 // CHECK1-NEXT: ret i32 [[TMP56]] 312 // 313 // 314 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28 315 // CHECK1-SAME: (i64 noundef [[TE:%.*]], i64 noundef [[TH:%.*]], i64 noundef [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1:[0-9]+]] { 316 // CHECK1-NEXT: entry: 317 // CHECK1-NEXT: [[TE_ADDR:%.*]] = alloca i64, align 8 318 // CHECK1-NEXT: [[TH_ADDR:%.*]] = alloca i64, align 8 319 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 320 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 8 321 // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB3]]) 322 // CHECK1-NEXT: store i64 [[TE]], i64* [[TE_ADDR]], align 8 323 // CHECK1-NEXT: store i64 [[TH]], i64* [[TH_ADDR]], align 8 324 // CHECK1-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 325 // CHECK1-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 8 326 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[TE_ADDR]] to i32* 327 // CHECK1-NEXT: [[CONV1:%.*]] = bitcast i64* [[TH_ADDR]] to i32* 328 // CHECK1-NEXT: [[CONV2:%.*]] = bitcast i64* [[N_ADDR]] to i32* 329 // CHECK1-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 8 330 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 331 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[CONV1]], align 4 332 // CHECK1-NEXT: call void @__kmpc_push_num_teams(%struct.ident_t* @[[GLOB3]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 333 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, [100 x i32]*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[CONV2]], [100 x i32]* [[TMP1]]) 334 // CHECK1-NEXT: ret void 335 // 336 // 337 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined. 338 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 339 // CHECK1-NEXT: entry: 340 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 341 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 342 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 343 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 8 344 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 345 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 346 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 347 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 348 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 349 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 350 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 351 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 352 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 353 // CHECK1-NEXT: [[I3:%.*]] = alloca i32, align 4 354 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 355 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 356 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 357 // CHECK1-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 8 358 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 359 // CHECK1-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 8 360 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP0]], align 4 361 // CHECK1-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 362 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 363 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 364 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 365 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 366 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 367 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 368 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 369 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 370 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 371 // CHECK1: omp.precond.then: 372 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 373 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 374 // CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_COMB_UB]], align 4 375 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 376 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 377 // CHECK1-NEXT: [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 378 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4 379 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP7]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 380 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 381 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 382 // CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 383 // CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 384 // CHECK1: cond.true: 385 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 386 // CHECK1-NEXT: br label [[COND_END:%.*]] 387 // CHECK1: cond.false: 388 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 389 // CHECK1-NEXT: br label [[COND_END]] 390 // CHECK1: cond.end: 391 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 392 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 393 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 394 // CHECK1-NEXT: store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4 395 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 396 // CHECK1: omp.inner.for.cond: 397 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 398 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 399 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 400 // CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 401 // CHECK1: omp.inner.for.body: 402 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 403 // CHECK1-NEXT: [[TMP16:%.*]] = zext i32 [[TMP15]] to i64 404 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 405 // CHECK1-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 406 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, [100 x i32]*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP16]], i64 [[TMP18]], i32* [[TMP0]], [100 x i32]* [[TMP1]]) 407 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 408 // CHECK1: omp.inner.for.inc: 409 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 410 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 411 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 412 // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 413 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 414 // CHECK1: omp.inner.for.end: 415 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 416 // CHECK1: omp.loop.exit: 417 // CHECK1-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 418 // CHECK1-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 419 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 420 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 421 // CHECK1: omp.precond.end: 422 // CHECK1-NEXT: ret void 423 // 424 // 425 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..1 426 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 427 // CHECK1-NEXT: entry: 428 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 429 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 430 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 431 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 432 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 433 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 8 434 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 435 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 436 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 437 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 438 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 439 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 440 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 441 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 442 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 443 // CHECK1-NEXT: [[I4:%.*]] = alloca i32, align 4 444 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 445 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 446 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 447 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 448 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 449 // CHECK1-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 8 450 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 451 // CHECK1-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 8 452 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP0]], align 4 453 // CHECK1-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 454 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 455 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 456 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 457 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 458 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 459 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 460 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 461 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 462 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 463 // CHECK1: omp.precond.then: 464 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 465 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 466 // CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4 467 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 468 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP6]] to i32 469 // CHECK1-NEXT: [[TMP7:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 470 // CHECK1-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP7]] to i32 471 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 472 // CHECK1-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 473 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 474 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 475 // CHECK1-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 476 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 477 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP9]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 478 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 479 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 480 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 481 // CHECK1-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 482 // CHECK1: cond.true: 483 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 484 // CHECK1-NEXT: br label [[COND_END:%.*]] 485 // CHECK1: cond.false: 486 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 487 // CHECK1-NEXT: br label [[COND_END]] 488 // CHECK1: cond.end: 489 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 490 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 491 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 492 // CHECK1-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 493 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 494 // CHECK1: omp.inner.for.cond: 495 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 496 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 497 // CHECK1-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 498 // CHECK1-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 499 // CHECK1: omp.inner.for.body: 500 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 501 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 502 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 503 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I4]], align 4 504 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[I4]], align 4 505 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP18]] to i64 506 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* [[TMP1]], i64 0, i64 [[IDXPROM]] 507 // CHECK1-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 508 // CHECK1-NEXT: [[TMP19:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 509 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4 510 // CHECK1-NEXT: [[TMP21:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* @[[GLOB3]], i32 [[TMP20]], i32 2) 511 // CHECK1-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 512 // CHECK1-NEXT: br i1 [[TMP22]], label [[DOTCANCEL_EXIT:%.*]], label [[DOTCANCEL_CONTINUE:%.*]] 513 // CHECK1: .cancel.exit: 514 // CHECK1-NEXT: br label [[CANCEL_EXIT:%.*]] 515 // CHECK1: .cancel.continue: 516 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 517 // CHECK1: omp.body.continue: 518 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 519 // CHECK1: omp.inner.for.inc: 520 // CHECK1-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 521 // CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP23]], 1 522 // CHECK1-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4 523 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 524 // CHECK1: omp.inner.for.end: 525 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 526 // CHECK1: omp.loop.exit: 527 // CHECK1-NEXT: [[TMP24:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 528 // CHECK1-NEXT: [[TMP25:%.*]] = load i32, i32* [[TMP24]], align 4 529 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP25]]) 530 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 531 // CHECK1: cancel.exit: 532 // CHECK1-NEXT: [[TMP26:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 533 // CHECK1-NEXT: [[TMP27:%.*]] = load i32, i32* [[TMP26]], align 4 534 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP27]]) 535 // CHECK1-NEXT: br label [[CANCEL_CONT:%.*]] 536 // CHECK1: omp.precond.end: 537 // CHECK1-NEXT: br label [[CANCEL_CONT]] 538 // CHECK1: cancel.cont: 539 // CHECK1-NEXT: ret void 540 // 541 // 542 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35 543 // CHECK1-SAME: (i64 noundef [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 544 // CHECK1-NEXT: entry: 545 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 546 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 8 547 // CHECK1-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 548 // CHECK1-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 8 549 // CHECK1-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 550 // CHECK1-NEXT: [[TMP0:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 8 551 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, [100 x i32]*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i32* [[CONV]], [100 x i32]* [[TMP0]]) 552 // CHECK1-NEXT: ret void 553 // 554 // 555 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..2 556 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 557 // CHECK1-NEXT: entry: 558 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 559 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 560 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 561 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 8 562 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 563 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 564 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 565 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 566 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 567 // CHECK1-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 568 // CHECK1-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 569 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 570 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 571 // CHECK1-NEXT: [[I3:%.*]] = alloca i32, align 4 572 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 573 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 574 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 575 // CHECK1-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 8 576 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 577 // CHECK1-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 8 578 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP0]], align 4 579 // CHECK1-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 580 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 581 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 582 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 583 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 584 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 585 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 586 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 587 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 588 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 589 // CHECK1: omp.precond.then: 590 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 591 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 592 // CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_COMB_UB]], align 4 593 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 594 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 595 // CHECK1-NEXT: [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 596 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4 597 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 598 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 599 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 600 // CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 601 // CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 602 // CHECK1: cond.true: 603 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 604 // CHECK1-NEXT: br label [[COND_END:%.*]] 605 // CHECK1: cond.false: 606 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 607 // CHECK1-NEXT: br label [[COND_END]] 608 // CHECK1: cond.end: 609 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 610 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 611 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 612 // CHECK1-NEXT: store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4 613 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 614 // CHECK1: omp.inner.for.cond: 615 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 616 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 617 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 618 // CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 619 // CHECK1: omp.inner.for.body: 620 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 621 // CHECK1-NEXT: [[TMP16:%.*]] = zext i32 [[TMP15]] to i64 622 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 623 // CHECK1-NEXT: [[TMP18:%.*]] = zext i32 [[TMP17]] to i64 624 // CHECK1-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, [100 x i32]*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP16]], i64 [[TMP18]], i32* [[TMP0]], [100 x i32]* [[TMP1]]) 625 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 626 // CHECK1: omp.inner.for.inc: 627 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 628 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 629 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]] 630 // CHECK1-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 631 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 632 // CHECK1: omp.inner.for.end: 633 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 634 // CHECK1: omp.loop.exit: 635 // CHECK1-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 636 // CHECK1-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 637 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 638 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 639 // CHECK1: omp.precond.end: 640 // CHECK1-NEXT: ret void 641 // 642 // 643 // CHECK1-LABEL: define {{[^@]+}}@.omp_outlined..3 644 // CHECK1-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 645 // CHECK1-NEXT: entry: 646 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 647 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 648 // CHECK1-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 649 // CHECK1-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 650 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 651 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 8 652 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 653 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 654 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 655 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 656 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 657 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 658 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 659 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 660 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 661 // CHECK1-NEXT: [[I4:%.*]] = alloca i32, align 4 662 // CHECK1-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 663 // CHECK1-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 664 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 665 // CHECK1-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 666 // CHECK1-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 667 // CHECK1-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 8 668 // CHECK1-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 669 // CHECK1-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 8 670 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP0]], align 4 671 // CHECK1-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 672 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 673 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 674 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 675 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 676 // CHECK1-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 677 // CHECK1-NEXT: store i32 0, i32* [[I]], align 4 678 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 679 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 680 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 681 // CHECK1: omp.precond.then: 682 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 683 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 684 // CHECK1-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4 685 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 686 // CHECK1-NEXT: [[CONV:%.*]] = trunc i64 [[TMP6]] to i32 687 // CHECK1-NEXT: [[TMP7:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 688 // CHECK1-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP7]] to i32 689 // CHECK1-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 690 // CHECK1-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 691 // CHECK1-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 692 // CHECK1-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 693 // CHECK1-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 694 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 695 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP9]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 696 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 697 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 698 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 699 // CHECK1-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 700 // CHECK1: cond.true: 701 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 702 // CHECK1-NEXT: br label [[COND_END:%.*]] 703 // CHECK1: cond.false: 704 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 705 // CHECK1-NEXT: br label [[COND_END]] 706 // CHECK1: cond.end: 707 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 708 // CHECK1-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 709 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 710 // CHECK1-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 711 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 712 // CHECK1: omp.inner.for.cond: 713 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 714 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 715 // CHECK1-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 716 // CHECK1-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 717 // CHECK1: omp.inner.for.body: 718 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 719 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 720 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 721 // CHECK1-NEXT: store i32 [[ADD]], i32* [[I4]], align 4 722 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, i32* [[I4]], align 4 723 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP18]] to i64 724 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* [[TMP1]], i64 0, i64 [[IDXPROM]] 725 // CHECK1-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 726 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 727 // CHECK1: omp.body.continue: 728 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 729 // CHECK1: omp.inner.for.inc: 730 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 731 // CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP19]], 1 732 // CHECK1-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4 733 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]] 734 // CHECK1: omp.inner.for.end: 735 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 736 // CHECK1: omp.loop.exit: 737 // CHECK1-NEXT: [[TMP20:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 738 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, i32* [[TMP20]], align 4 739 // CHECK1-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP21]]) 740 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 741 // CHECK1: omp.precond.end: 742 // CHECK1-NEXT: ret void 743 // 744 // 745 // CHECK1-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 746 // CHECK1-SAME: () #[[ATTR3:[0-9]+]] { 747 // CHECK1-NEXT: entry: 748 // CHECK1-NEXT: call void @__tgt_register_requires(i64 1) 749 // CHECK1-NEXT: ret void 750 // 751 // 752 // CHECK3-LABEL: define {{[^@]+}}@_Z21teams_argument_globali 753 // CHECK3-SAME: (i32 noundef [[N:%.*]]) #[[ATTR0:[0-9]+]] { 754 // CHECK3-NEXT: entry: 755 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 756 // CHECK3-NEXT: [[TE:%.*]] = alloca i32, align 4 757 // CHECK3-NEXT: [[TH:%.*]] = alloca i32, align 4 758 // CHECK3-NEXT: [[TE_CASTED:%.*]] = alloca i32, align 4 759 // CHECK3-NEXT: [[TH_CASTED:%.*]] = alloca i32, align 4 760 // CHECK3-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 761 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x i8*], align 4 762 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x i8*], align 4 763 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x i8*], align 4 764 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 765 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 766 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 767 // CHECK3-NEXT: [[N_CASTED4:%.*]] = alloca i32, align 4 768 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [2 x i8*], align 4 769 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [2 x i8*], align 4 770 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [2 x i8*], align 4 771 // CHECK3-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 772 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 773 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 774 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 775 // CHECK3-NEXT: [[TMP0:%.*]] = load i32, i32* [[N_ADDR]], align 4 776 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP0]], 128 777 // CHECK3-NEXT: store i32 [[DIV]], i32* [[TE]], align 4 778 // CHECK3-NEXT: store i32 128, i32* [[TH]], align 4 779 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, i32* [[TE]], align 4 780 // CHECK3-NEXT: store i32 [[TMP1]], i32* [[TE_CASTED]], align 4 781 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[TE_CASTED]], align 4 782 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[TH]], align 4 783 // CHECK3-NEXT: store i32 [[TMP3]], i32* [[TH_CASTED]], align 4 784 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[TH_CASTED]], align 4 785 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[N_ADDR]], align 4 786 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[N_CASTED]], align 4 787 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[N_CASTED]], align 4 788 // CHECK3-NEXT: [[TMP7:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 789 // CHECK3-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i32* 790 // CHECK3-NEXT: store i32 [[TMP2]], i32* [[TMP8]], align 4 791 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 792 // CHECK3-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32* 793 // CHECK3-NEXT: store i32 [[TMP2]], i32* [[TMP10]], align 4 794 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 795 // CHECK3-NEXT: store i8* null, i8** [[TMP11]], align 4 796 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 797 // CHECK3-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* 798 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[TMP13]], align 4 799 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 800 // CHECK3-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* 801 // CHECK3-NEXT: store i32 [[TMP4]], i32* [[TMP15]], align 4 802 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 803 // CHECK3-NEXT: store i8* null, i8** [[TMP16]], align 4 804 // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 805 // CHECK3-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32* 806 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[TMP18]], align 4 807 // CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 808 // CHECK3-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32* 809 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[TMP20]], align 4 810 // CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 811 // CHECK3-NEXT: store i8* null, i8** [[TMP21]], align 4 812 // CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 813 // CHECK3-NEXT: [[TMP23:%.*]] = bitcast i8** [[TMP22]] to [100 x i32]** 814 // CHECK3-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP23]], align 4 815 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 3 816 // CHECK3-NEXT: [[TMP25:%.*]] = bitcast i8** [[TMP24]] to [100 x i32]** 817 // CHECK3-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP25]], align 4 818 // CHECK3-NEXT: [[TMP26:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 819 // CHECK3-NEXT: store i8* null, i8** [[TMP26]], align 4 820 // CHECK3-NEXT: [[TMP27:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 821 // CHECK3-NEXT: [[TMP28:%.*]] = getelementptr inbounds [4 x i8*], [4 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 822 // CHECK3-NEXT: [[TMP29:%.*]] = load i32, i32* [[TE]], align 4 823 // CHECK3-NEXT: [[TMP30:%.*]] = load i32, i32* [[N_ADDR]], align 4 824 // CHECK3-NEXT: store i32 [[TMP30]], i32* [[DOTCAPTURE_EXPR_]], align 4 825 // CHECK3-NEXT: [[TMP31:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 826 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP31]], 0 827 // CHECK3-NEXT: [[DIV2:%.*]] = sdiv i32 [[SUB]], 1 828 // CHECK3-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV2]], 1 829 // CHECK3-NEXT: store i32 [[SUB3]], i32* [[DOTCAPTURE_EXPR_1]], align 4 830 // CHECK3-NEXT: [[TMP32:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 831 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP32]], 1 832 // CHECK3-NEXT: [[TMP33:%.*]] = zext i32 [[ADD]] to i64 833 // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP33]]) 834 // CHECK3-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28.region_id, i32 4, i8** [[TMP27]], i8** [[TMP28]], i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([4 x i64], [4 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 [[TMP29]], i32 0) 835 // CHECK3-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0 836 // CHECK3-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 837 // CHECK3: omp_offload.failed: 838 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28(i32 [[TMP2]], i32 [[TMP4]], i32 [[TMP6]], [100 x i32]* @a) #[[ATTR2:[0-9]+]] 839 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] 840 // CHECK3: omp_offload.cont: 841 // CHECK3-NEXT: [[TMP36:%.*]] = load i32, i32* [[N_ADDR]], align 4 842 // CHECK3-NEXT: store i32 [[TMP36]], i32* [[N_CASTED4]], align 4 843 // CHECK3-NEXT: [[TMP37:%.*]] = load i32, i32* [[N_CASTED4]], align 4 844 // CHECK3-NEXT: [[TMP38:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 845 // CHECK3-NEXT: [[TMP39:%.*]] = bitcast i8** [[TMP38]] to i32* 846 // CHECK3-NEXT: store i32 [[TMP37]], i32* [[TMP39]], align 4 847 // CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 848 // CHECK3-NEXT: [[TMP41:%.*]] = bitcast i8** [[TMP40]] to i32* 849 // CHECK3-NEXT: store i32 [[TMP37]], i32* [[TMP41]], align 4 850 // CHECK3-NEXT: [[TMP42:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i32 0, i32 0 851 // CHECK3-NEXT: store i8* null, i8** [[TMP42]], align 4 852 // CHECK3-NEXT: [[TMP43:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 853 // CHECK3-NEXT: [[TMP44:%.*]] = bitcast i8** [[TMP43]] to [100 x i32]** 854 // CHECK3-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP44]], align 4 855 // CHECK3-NEXT: [[TMP45:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 856 // CHECK3-NEXT: [[TMP46:%.*]] = bitcast i8** [[TMP45]] to [100 x i32]** 857 // CHECK3-NEXT: store [100 x i32]* @a, [100 x i32]** [[TMP46]], align 4 858 // CHECK3-NEXT: [[TMP47:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_MAPPERS7]], i32 0, i32 1 859 // CHECK3-NEXT: store i8* null, i8** [[TMP47]], align 4 860 // CHECK3-NEXT: [[TMP48:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 861 // CHECK3-NEXT: [[TMP49:%.*]] = getelementptr inbounds [2 x i8*], [2 x i8*]* [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 862 // CHECK3-NEXT: [[TMP50:%.*]] = load i32, i32* [[N_ADDR]], align 4 863 // CHECK3-NEXT: store i32 [[TMP50]], i32* [[DOTCAPTURE_EXPR_9]], align 4 864 // CHECK3-NEXT: [[TMP51:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_9]], align 4 865 // CHECK3-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP51]], 0 866 // CHECK3-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 867 // CHECK3-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 868 // CHECK3-NEXT: store i32 [[SUB13]], i32* [[DOTCAPTURE_EXPR_10]], align 4 869 // CHECK3-NEXT: [[TMP52:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_10]], align 4 870 // CHECK3-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP52]], 1 871 // CHECK3-NEXT: [[TMP53:%.*]] = zext i32 [[ADD14]] to i64 872 // CHECK3-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 [[TMP53]]) 873 // CHECK3-NEXT: [[TMP54:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35.region_id, i32 2, i8** [[TMP48]], i8** [[TMP49]], i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([2 x i64], [2 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) 874 // CHECK3-NEXT: [[TMP55:%.*]] = icmp ne i32 [[TMP54]], 0 875 // CHECK3-NEXT: br i1 [[TMP55]], label [[OMP_OFFLOAD_FAILED15:%.*]], label [[OMP_OFFLOAD_CONT16:%.*]] 876 // CHECK3: omp_offload.failed15: 877 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35(i32 [[TMP37]], [100 x i32]* @a) #[[ATTR2]] 878 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT16]] 879 // CHECK3: omp_offload.cont16: 880 // CHECK3-NEXT: [[TMP56:%.*]] = load i32, i32* getelementptr inbounds ([100 x i32], [100 x i32]* @a, i32 0, i32 0), align 4 881 // CHECK3-NEXT: ret i32 [[TMP56]] 882 // 883 // 884 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l28 885 // CHECK3-SAME: (i32 noundef [[TE:%.*]], i32 noundef [[TH:%.*]], i32 noundef [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1:[0-9]+]] { 886 // CHECK3-NEXT: entry: 887 // CHECK3-NEXT: [[TE_ADDR:%.*]] = alloca i32, align 4 888 // CHECK3-NEXT: [[TH_ADDR:%.*]] = alloca i32, align 4 889 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 890 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 4 891 // CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB3]]) 892 // CHECK3-NEXT: store i32 [[TE]], i32* [[TE_ADDR]], align 4 893 // CHECK3-NEXT: store i32 [[TH]], i32* [[TH_ADDR]], align 4 894 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 895 // CHECK3-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 4 896 // CHECK3-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 4 897 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[TE_ADDR]], align 4 898 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[TH_ADDR]], align 4 899 // CHECK3-NEXT: call void @__kmpc_push_num_teams(%struct.ident_t* @[[GLOB3]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 900 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, [100 x i32]*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[N_ADDR]], [100 x i32]* [[TMP1]]) 901 // CHECK3-NEXT: ret void 902 // 903 // 904 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined. 905 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 906 // CHECK3-NEXT: entry: 907 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 908 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 909 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 910 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 4 911 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 912 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 913 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 914 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 915 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 916 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 917 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 918 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 919 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 920 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 921 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 922 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 923 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 924 // CHECK3-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 4 925 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 926 // CHECK3-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 4 927 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP0]], align 4 928 // CHECK3-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 929 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 930 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 931 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 932 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 933 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 934 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 935 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 936 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 937 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 938 // CHECK3: omp.precond.then: 939 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 940 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 941 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_COMB_UB]], align 4 942 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 943 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 944 // CHECK3-NEXT: [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 945 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4 946 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP7]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 947 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 948 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 949 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 950 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 951 // CHECK3: cond.true: 952 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 953 // CHECK3-NEXT: br label [[COND_END:%.*]] 954 // CHECK3: cond.false: 955 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 956 // CHECK3-NEXT: br label [[COND_END]] 957 // CHECK3: cond.end: 958 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 959 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 960 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 961 // CHECK3-NEXT: store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4 962 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 963 // CHECK3: omp.inner.for.cond: 964 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 965 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 966 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 967 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 968 // CHECK3: omp.inner.for.body: 969 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 970 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 971 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, [100 x i32]*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32 [[TMP15]], i32 [[TMP16]], i32* [[TMP0]], [100 x i32]* [[TMP1]]) 972 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 973 // CHECK3: omp.inner.for.inc: 974 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 975 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 976 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP17]], [[TMP18]] 977 // CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 978 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 979 // CHECK3: omp.inner.for.end: 980 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 981 // CHECK3: omp.loop.exit: 982 // CHECK3-NEXT: [[TMP19:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 983 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4 984 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP20]]) 985 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 986 // CHECK3: omp.precond.end: 987 // CHECK3-NEXT: ret void 988 // 989 // 990 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..1 991 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 992 // CHECK3-NEXT: entry: 993 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 994 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 995 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 996 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 997 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 998 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 4 999 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1000 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 1001 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1002 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1003 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 1004 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1005 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1006 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1007 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1008 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 1009 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 1010 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 1011 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 1012 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 1013 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 1014 // CHECK3-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 4 1015 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 1016 // CHECK3-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 4 1017 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP0]], align 4 1018 // CHECK3-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 1019 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1020 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 1021 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1022 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1023 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1024 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 1025 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1026 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 1027 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1028 // CHECK3: omp.precond.then: 1029 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1030 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1031 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4 1032 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 1033 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 1034 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_LB]], align 4 1035 // CHECK3-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 1036 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1037 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1038 // CHECK3-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 1039 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 1040 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP9]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1041 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1042 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1043 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 1044 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1045 // CHECK3: cond.true: 1046 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1047 // CHECK3-NEXT: br label [[COND_END:%.*]] 1048 // CHECK3: cond.false: 1049 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1050 // CHECK3-NEXT: br label [[COND_END]] 1051 // CHECK3: cond.end: 1052 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 1053 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 1054 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1055 // CHECK3-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 1056 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1057 // CHECK3: omp.inner.for.cond: 1058 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1059 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1060 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 1061 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1062 // CHECK3: omp.inner.for.body: 1063 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1064 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 1065 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1066 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I3]], align 4 1067 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[I3]], align 4 1068 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* [[TMP1]], i32 0, i32 [[TMP18]] 1069 // CHECK3-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 1070 // CHECK3-NEXT: [[TMP19:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 1071 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4 1072 // CHECK3-NEXT: [[TMP21:%.*]] = call i32 @__kmpc_cancel(%struct.ident_t* @[[GLOB3]], i32 [[TMP20]], i32 2) 1073 // CHECK3-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 1074 // CHECK3-NEXT: br i1 [[TMP22]], label [[DOTCANCEL_EXIT:%.*]], label [[DOTCANCEL_CONTINUE:%.*]] 1075 // CHECK3: .cancel.exit: 1076 // CHECK3-NEXT: br label [[CANCEL_EXIT:%.*]] 1077 // CHECK3: .cancel.continue: 1078 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1079 // CHECK3: omp.body.continue: 1080 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1081 // CHECK3: omp.inner.for.inc: 1082 // CHECK3-NEXT: [[TMP23:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1083 // CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP23]], 1 1084 // CHECK3-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4 1085 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 1086 // CHECK3: omp.inner.for.end: 1087 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1088 // CHECK3: omp.loop.exit: 1089 // CHECK3-NEXT: [[TMP24:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 1090 // CHECK3-NEXT: [[TMP25:%.*]] = load i32, i32* [[TMP24]], align 4 1091 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP25]]) 1092 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 1093 // CHECK3: cancel.exit: 1094 // CHECK3-NEXT: [[TMP26:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 1095 // CHECK3-NEXT: [[TMP27:%.*]] = load i32, i32* [[TMP26]], align 4 1096 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP27]]) 1097 // CHECK3-NEXT: br label [[CANCEL_CONT:%.*]] 1098 // CHECK3: omp.precond.end: 1099 // CHECK3-NEXT: br label [[CANCEL_CONT]] 1100 // CHECK3: cancel.cont: 1101 // CHECK3-NEXT: ret void 1102 // 1103 // 1104 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l35 1105 // CHECK3-SAME: (i32 noundef [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 1106 // CHECK3-NEXT: entry: 1107 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 1108 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 4 1109 // CHECK3-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 1110 // CHECK3-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 4 1111 // CHECK3-NEXT: [[TMP0:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 4 1112 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 2, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, [100 x i32]*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), i32* [[N_ADDR]], [100 x i32]* [[TMP0]]) 1113 // CHECK3-NEXT: ret void 1114 // 1115 // 1116 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..2 1117 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 1118 // CHECK3-NEXT: entry: 1119 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 1120 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 1121 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 1122 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 4 1123 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1124 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 1125 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1126 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1127 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 1128 // CHECK3-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 1129 // CHECK3-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 1130 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1131 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1132 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 1133 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 1134 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 1135 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 1136 // CHECK3-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 4 1137 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 1138 // CHECK3-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 4 1139 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP0]], align 4 1140 // CHECK3-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 1141 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1142 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 1143 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1144 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1145 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1146 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 1147 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1148 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 1149 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1150 // CHECK3: omp.precond.then: 1151 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 1152 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1153 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_COMB_UB]], align 4 1154 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1155 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1156 // CHECK3-NEXT: [[TMP6:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 1157 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[TMP6]], align 4 1158 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP7]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1159 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1160 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1161 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 1162 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1163 // CHECK3: cond.true: 1164 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1165 // CHECK3-NEXT: br label [[COND_END:%.*]] 1166 // CHECK3: cond.false: 1167 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1168 // CHECK3-NEXT: br label [[COND_END]] 1169 // CHECK3: cond.end: 1170 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 1171 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 1172 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 1173 // CHECK3-NEXT: store i32 [[TMP12]], i32* [[DOTOMP_IV]], align 4 1174 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1175 // CHECK3: omp.inner.for.cond: 1176 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1177 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1178 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 1179 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1180 // CHECK3: omp.inner.for.body: 1181 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 1182 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1183 // CHECK3-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 4, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, [100 x i32]*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32 [[TMP15]], i32 [[TMP16]], i32* [[TMP0]], [100 x i32]* [[TMP1]]) 1184 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1185 // CHECK3: omp.inner.for.inc: 1186 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1187 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 1188 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP17]], [[TMP18]] 1189 // CHECK3-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 1190 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 1191 // CHECK3: omp.inner.for.end: 1192 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1193 // CHECK3: omp.loop.exit: 1194 // CHECK3-NEXT: [[TMP19:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 1195 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, i32* [[TMP19]], align 4 1196 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP20]]) 1197 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 1198 // CHECK3: omp.precond.end: 1199 // CHECK3-NEXT: ret void 1200 // 1201 // 1202 // CHECK3-LABEL: define {{[^@]+}}@.omp_outlined..3 1203 // CHECK3-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], [100 x i32]* noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 1204 // CHECK3-NEXT: entry: 1205 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 1206 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 1207 // CHECK3-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 1208 // CHECK3-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 1209 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 1210 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca [100 x i32]*, align 4 1211 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1212 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 1213 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1214 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1215 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 1216 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1217 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1218 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1219 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1220 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 1221 // CHECK3-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 1222 // CHECK3-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 1223 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 1224 // CHECK3-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 1225 // CHECK3-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 1226 // CHECK3-NEXT: store [100 x i32]* [[A]], [100 x i32]** [[A_ADDR]], align 4 1227 // CHECK3-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 1228 // CHECK3-NEXT: [[TMP1:%.*]] = load [100 x i32]*, [100 x i32]** [[A_ADDR]], align 4 1229 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP0]], align 4 1230 // CHECK3-NEXT: store i32 [[TMP2]], i32* [[DOTCAPTURE_EXPR_]], align 4 1231 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1232 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 1233 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1234 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1235 // CHECK3-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1236 // CHECK3-NEXT: store i32 0, i32* [[I]], align 4 1237 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1238 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 1239 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1240 // CHECK3: omp.precond.then: 1241 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1242 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1243 // CHECK3-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_UB]], align 4 1244 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 1245 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 1246 // CHECK3-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_LB]], align 4 1247 // CHECK3-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_UB]], align 4 1248 // CHECK3-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1249 // CHECK3-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1250 // CHECK3-NEXT: [[TMP8:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 1251 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, i32* [[TMP8]], align 4 1252 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP9]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1253 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1254 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1255 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP10]], [[TMP11]] 1256 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1257 // CHECK3: cond.true: 1258 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1259 // CHECK3-NEXT: br label [[COND_END:%.*]] 1260 // CHECK3: cond.false: 1261 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1262 // CHECK3-NEXT: br label [[COND_END]] 1263 // CHECK3: cond.end: 1264 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP12]], [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ] 1265 // CHECK3-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 1266 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1267 // CHECK3-NEXT: store i32 [[TMP14]], i32* [[DOTOMP_IV]], align 4 1268 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1269 // CHECK3: omp.inner.for.cond: 1270 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1271 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1272 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 1273 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1274 // CHECK3: omp.inner.for.body: 1275 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1276 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 1277 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1278 // CHECK3-NEXT: store i32 [[ADD]], i32* [[I3]], align 4 1279 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, i32* [[I3]], align 4 1280 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], [100 x i32]* [[TMP1]], i32 0, i32 [[TMP18]] 1281 // CHECK3-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 1282 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1283 // CHECK3: omp.body.continue: 1284 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1285 // CHECK3: omp.inner.for.inc: 1286 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1287 // CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP19]], 1 1288 // CHECK3-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4 1289 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]] 1290 // CHECK3: omp.inner.for.end: 1291 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1292 // CHECK3: omp.loop.exit: 1293 // CHECK3-NEXT: [[TMP20:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 1294 // CHECK3-NEXT: [[TMP21:%.*]] = load i32, i32* [[TMP20]], align 4 1295 // CHECK3-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP21]]) 1296 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 1297 // CHECK3: omp.precond.end: 1298 // CHECK3-NEXT: ret void 1299 // 1300 // 1301 // CHECK3-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 1302 // CHECK3-SAME: () #[[ATTR3:[0-9]+]] { 1303 // CHECK3-NEXT: entry: 1304 // CHECK3-NEXT: call void @__tgt_register_requires(i64 1) 1305 // CHECK3-NEXT: ret void 1306 // 1307 // 1308 // CHECK9-LABEL: define {{[^@]+}}@_Z15teams_local_argv 1309 // CHECK9-SAME: () #[[ATTR0:[0-9]+]] { 1310 // CHECK9-NEXT: entry: 1311 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 1312 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 1313 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 1314 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 1315 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 1316 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 1317 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 1318 // CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8 1319 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1320 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1321 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1322 // CHECK9-NEXT: store i32 100, i32* [[N]], align 4 1323 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 1324 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 1325 // CHECK9-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() 1326 // CHECK9-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 1327 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 1328 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 1329 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 1330 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* 1331 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 1332 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 1333 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 1334 // CHECK9-NEXT: [[TMP6:%.*]] = bitcast [3 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 1335 // CHECK9-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP6]], i8* align 8 bitcast ([3 x i64]* @.offload_sizes to i8*), i64 24, i1 false) 1336 // CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1337 // CHECK9-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64* 1338 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP8]], align 8 1339 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1340 // CHECK9-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64* 1341 // CHECK9-NEXT: store i64 [[TMP4]], i64* [[TMP10]], align 8 1342 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 1343 // CHECK9-NEXT: store i8* null, i8** [[TMP11]], align 8 1344 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 1345 // CHECK9-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* 1346 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 1347 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 1348 // CHECK9-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* 1349 // CHECK9-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 1350 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 1351 // CHECK9-NEXT: store i8* null, i8** [[TMP16]], align 8 1352 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 1353 // CHECK9-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** 1354 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP18]], align 8 1355 // CHECK9-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 1356 // CHECK9-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32** 1357 // CHECK9-NEXT: store i32* [[VLA]], i32** [[TMP20]], align 8 1358 // CHECK9-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 1359 // CHECK9-NEXT: store i64 [[TMP5]], i64* [[TMP21]], align 8 1360 // CHECK9-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 1361 // CHECK9-NEXT: store i8* null, i8** [[TMP22]], align 8 1362 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1363 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1364 // CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 1365 // CHECK9-NEXT: [[TMP26:%.*]] = load i32, i32* [[N]], align 4 1366 // CHECK9-NEXT: store i32 [[TMP26]], i32* [[DOTCAPTURE_EXPR_]], align 4 1367 // CHECK9-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1368 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP27]], 0 1369 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1370 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1371 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1372 // CHECK9-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1373 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP28]], 1 1374 // CHECK9-NEXT: [[TMP29:%.*]] = zext i32 [[ADD]] to i64 1375 // CHECK9-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP29]]) 1376 // CHECK9-NEXT: [[TMP30:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP23]], i8** [[TMP24]], i64* [[TMP25]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) 1377 // CHECK9-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0 1378 // CHECK9-NEXT: br i1 [[TMP31]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1379 // CHECK9: omp_offload.failed: 1380 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] 1381 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] 1382 // CHECK9: omp_offload.cont: 1383 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i64 0 1384 // CHECK9-NEXT: [[TMP32:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 1385 // CHECK9-NEXT: [[TMP33:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 1386 // CHECK9-NEXT: call void @llvm.stackrestore(i8* [[TMP33]]) 1387 // CHECK9-NEXT: ret i32 [[TMP32]] 1388 // 1389 // 1390 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73 1391 // CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 1392 // CHECK9-NEXT: entry: 1393 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 1394 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1395 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 1396 // CHECK9-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 1397 // CHECK9-NEXT: store i64 [[VLA]], i64* [[VLA_ADDR]], align 8 1398 // CHECK9-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 1399 // CHECK9-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 1400 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 1401 // CHECK9-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 1402 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[CONV]], i64 [[TMP0]], i32* [[TMP1]]) 1403 // CHECK9-NEXT: ret void 1404 // 1405 // 1406 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined. 1407 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1408 // CHECK9-NEXT: entry: 1409 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1410 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1411 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 1412 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1413 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 1414 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1415 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1416 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1417 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1418 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 1419 // CHECK9-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 1420 // CHECK9-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 1421 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1422 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1423 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 1424 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1425 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1426 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 1427 // CHECK9-NEXT: store i64 [[VLA]], i64* [[VLA_ADDR]], align 8 1428 // CHECK9-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 1429 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 1430 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 1431 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 8 1432 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 1433 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 1434 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1435 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1436 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1437 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1438 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1439 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 1440 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1441 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1442 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1443 // CHECK9: omp.precond.then: 1444 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 1445 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1446 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_COMB_UB]], align 4 1447 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1448 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1449 // CHECK9-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1450 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 1451 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1452 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1453 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1454 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 1455 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1456 // CHECK9: cond.true: 1457 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1458 // CHECK9-NEXT: br label [[COND_END:%.*]] 1459 // CHECK9: cond.false: 1460 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1461 // CHECK9-NEXT: br label [[COND_END]] 1462 // CHECK9: cond.end: 1463 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 1464 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 1465 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 1466 // CHECK9-NEXT: store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4 1467 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1468 // CHECK9: omp.inner.for.cond: 1469 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1470 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1471 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 1472 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1473 // CHECK9: omp.inner.for.body: 1474 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 1475 // CHECK9-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64 1476 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1477 // CHECK9-NEXT: [[TMP19:%.*]] = zext i32 [[TMP18]] to i64 1478 // CHECK9-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP17]], i64 [[TMP19]], i32* [[TMP0]], i64 [[TMP1]], i32* [[TMP2]]) 1479 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1480 // CHECK9: omp.inner.for.inc: 1481 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1482 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 1483 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]] 1484 // CHECK9-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 1485 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 1486 // CHECK9: omp.inner.for.end: 1487 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1488 // CHECK9: omp.loop.exit: 1489 // CHECK9-NEXT: [[TMP22:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1490 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, i32* [[TMP22]], align 4 1491 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP23]]) 1492 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 1493 // CHECK9: omp.precond.end: 1494 // CHECK9-NEXT: ret void 1495 // 1496 // 1497 // CHECK9-LABEL: define {{[^@]+}}@.omp_outlined..1 1498 // CHECK9-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1499 // CHECK9-NEXT: entry: 1500 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1501 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1502 // CHECK9-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 1503 // CHECK9-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 1504 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 1505 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1506 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 1507 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1508 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1509 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1510 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1511 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 1512 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1513 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1514 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1515 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1516 // CHECK9-NEXT: [[I4:%.*]] = alloca i32, align 4 1517 // CHECK9-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1518 // CHECK9-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1519 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 1520 // CHECK9-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 1521 // CHECK9-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 1522 // CHECK9-NEXT: store i64 [[VLA]], i64* [[VLA_ADDR]], align 8 1523 // CHECK9-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 1524 // CHECK9-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 1525 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 1526 // CHECK9-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 8 1527 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 1528 // CHECK9-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 1529 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1530 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1531 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1532 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1533 // CHECK9-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1534 // CHECK9-NEXT: store i32 0, i32* [[I]], align 4 1535 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1536 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1537 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1538 // CHECK9: omp.precond.then: 1539 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 1540 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1541 // CHECK9-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_UB]], align 4 1542 // CHECK9-NEXT: [[TMP7:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 1543 // CHECK9-NEXT: [[CONV:%.*]] = trunc i64 [[TMP7]] to i32 1544 // CHECK9-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 1545 // CHECK9-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP8]] to i32 1546 // CHECK9-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 1547 // CHECK9-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 1548 // CHECK9-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1549 // CHECK9-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1550 // CHECK9-NEXT: [[TMP9:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1551 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4 1552 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1553 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1554 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1555 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]] 1556 // CHECK9-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1557 // CHECK9: cond.true: 1558 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1559 // CHECK9-NEXT: br label [[COND_END:%.*]] 1560 // CHECK9: cond.false: 1561 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1562 // CHECK9-NEXT: br label [[COND_END]] 1563 // CHECK9: cond.end: 1564 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ] 1565 // CHECK9-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 1566 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1567 // CHECK9-NEXT: store i32 [[TMP15]], i32* [[DOTOMP_IV]], align 4 1568 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1569 // CHECK9: omp.inner.for.cond: 1570 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1571 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1572 // CHECK9-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]] 1573 // CHECK9-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1574 // CHECK9: omp.inner.for.body: 1575 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1576 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1 1577 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1578 // CHECK9-NEXT: store i32 [[ADD]], i32* [[I4]], align 4 1579 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, i32* [[I4]], align 4 1580 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP19]] to i64 1581 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i64 [[IDXPROM]] 1582 // CHECK9-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 1583 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1584 // CHECK9: omp.body.continue: 1585 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1586 // CHECK9: omp.inner.for.inc: 1587 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1588 // CHECK9-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP20]], 1 1589 // CHECK9-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4 1590 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]] 1591 // CHECK9: omp.inner.for.end: 1592 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1593 // CHECK9: omp.loop.exit: 1594 // CHECK9-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1595 // CHECK9-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 1596 // CHECK9-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 1597 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 1598 // CHECK9: omp.precond.end: 1599 // CHECK9-NEXT: ret void 1600 // 1601 // 1602 // CHECK9-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 1603 // CHECK9-SAME: () #[[ATTR5:[0-9]+]] { 1604 // CHECK9-NEXT: entry: 1605 // CHECK9-NEXT: call void @__tgt_register_requires(i64 1) 1606 // CHECK9-NEXT: ret void 1607 // 1608 // 1609 // CHECK11-LABEL: define {{[^@]+}}@_Z15teams_local_argv 1610 // CHECK11-SAME: () #[[ATTR0:[0-9]+]] { 1611 // CHECK11-NEXT: entry: 1612 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 1613 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 1614 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 1615 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 1616 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 1617 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 1618 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 1619 // CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4 1620 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 1621 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1622 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1623 // CHECK11-NEXT: store i32 100, i32* [[N]], align 4 1624 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 1625 // CHECK11-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() 1626 // CHECK11-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 1627 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 1628 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 1629 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 1630 // CHECK11-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 1631 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 1632 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 1633 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 1634 // CHECK11-NEXT: [[TMP6:%.*]] = bitcast [3 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 1635 // CHECK11-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP6]], i8* align 4 bitcast ([3 x i64]* @.offload_sizes to i8*), i32 24, i1 false) 1636 // CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1637 // CHECK11-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i32* 1638 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP8]], align 4 1639 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1640 // CHECK11-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32* 1641 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[TMP10]], align 4 1642 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 1643 // CHECK11-NEXT: store i8* null, i8** [[TMP11]], align 4 1644 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 1645 // CHECK11-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* 1646 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 1647 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 1648 // CHECK11-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* 1649 // CHECK11-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 1650 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 1651 // CHECK11-NEXT: store i8* null, i8** [[TMP16]], align 4 1652 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 1653 // CHECK11-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** 1654 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP18]], align 4 1655 // CHECK11-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 1656 // CHECK11-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32** 1657 // CHECK11-NEXT: store i32* [[VLA]], i32** [[TMP20]], align 4 1658 // CHECK11-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 1659 // CHECK11-NEXT: store i64 [[TMP5]], i64* [[TMP21]], align 4 1660 // CHECK11-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 1661 // CHECK11-NEXT: store i8* null, i8** [[TMP22]], align 4 1662 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1663 // CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1664 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 1665 // CHECK11-NEXT: [[TMP26:%.*]] = load i32, i32* [[N]], align 4 1666 // CHECK11-NEXT: store i32 [[TMP26]], i32* [[DOTCAPTURE_EXPR_]], align 4 1667 // CHECK11-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1668 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP27]], 0 1669 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1670 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1671 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1672 // CHECK11-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1673 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP28]], 1 1674 // CHECK11-NEXT: [[TMP29:%.*]] = zext i32 [[ADD]] to i64 1675 // CHECK11-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP29]]) 1676 // CHECK11-NEXT: [[TMP30:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73.region_id, i32 3, i8** [[TMP23]], i8** [[TMP24]], i64* [[TMP25]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) 1677 // CHECK11-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0 1678 // CHECK11-NEXT: br i1 [[TMP31]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1679 // CHECK11: omp_offload.failed: 1680 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] 1681 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] 1682 // CHECK11: omp_offload.cont: 1683 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[VLA]], i32 0 1684 // CHECK11-NEXT: [[TMP32:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 1685 // CHECK11-NEXT: [[TMP33:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 1686 // CHECK11-NEXT: call void @llvm.stackrestore(i8* [[TMP33]]) 1687 // CHECK11-NEXT: ret i32 [[TMP32]] 1688 // 1689 // 1690 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l73 1691 // CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 1692 // CHECK11-NEXT: entry: 1693 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 1694 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 1695 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 1696 // CHECK11-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 1697 // CHECK11-NEXT: store i32 [[VLA]], i32* [[VLA_ADDR]], align 4 1698 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 1699 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 1700 // CHECK11-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 1701 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]]) 1702 // CHECK11-NEXT: ret void 1703 // 1704 // 1705 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined. 1706 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1707 // CHECK11-NEXT: entry: 1708 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 1709 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 1710 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 1711 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 1712 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 1713 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1714 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 1715 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1716 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1717 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 1718 // CHECK11-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 1719 // CHECK11-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 1720 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1721 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1722 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 1723 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 1724 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 1725 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 1726 // CHECK11-NEXT: store i32 [[VLA]], i32* [[VLA_ADDR]], align 4 1727 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 1728 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 1729 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 1730 // CHECK11-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 4 1731 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 1732 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 1733 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1734 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1735 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1736 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1737 // CHECK11-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1738 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 1739 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 1740 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1741 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1742 // CHECK11: omp.precond.then: 1743 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 1744 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1745 // CHECK11-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_COMB_UB]], align 4 1746 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1747 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1748 // CHECK11-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 1749 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 1750 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1751 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1752 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1753 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 1754 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1755 // CHECK11: cond.true: 1756 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1757 // CHECK11-NEXT: br label [[COND_END:%.*]] 1758 // CHECK11: cond.false: 1759 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1760 // CHECK11-NEXT: br label [[COND_END]] 1761 // CHECK11: cond.end: 1762 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 1763 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 1764 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 1765 // CHECK11-NEXT: store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4 1766 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1767 // CHECK11: omp.inner.for.cond: 1768 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1769 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1770 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 1771 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1772 // CHECK11: omp.inner.for.body: 1773 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 1774 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1775 // CHECK11-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32 [[TMP16]], i32 [[TMP17]], i32* [[TMP0]], i32 [[TMP1]], i32* [[TMP2]]) 1776 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1777 // CHECK11: omp.inner.for.inc: 1778 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1779 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 1780 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] 1781 // CHECK11-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 1782 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 1783 // CHECK11: omp.inner.for.end: 1784 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1785 // CHECK11: omp.loop.exit: 1786 // CHECK11-NEXT: [[TMP20:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 1787 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, i32* [[TMP20]], align 4 1788 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP21]]) 1789 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 1790 // CHECK11: omp.precond.end: 1791 // CHECK11-NEXT: ret void 1792 // 1793 // 1794 // CHECK11-LABEL: define {{[^@]+}}@.omp_outlined..1 1795 // CHECK11-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1796 // CHECK11-NEXT: entry: 1797 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 1798 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 1799 // CHECK11-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 1800 // CHECK11-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 1801 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 1802 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 1803 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 1804 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1805 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 1806 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1807 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1808 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 1809 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1810 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1811 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1812 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1813 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 1814 // CHECK11-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 1815 // CHECK11-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 1816 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 1817 // CHECK11-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 1818 // CHECK11-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 1819 // CHECK11-NEXT: store i32 [[VLA]], i32* [[VLA_ADDR]], align 4 1820 // CHECK11-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 1821 // CHECK11-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 1822 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 1823 // CHECK11-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 4 1824 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 1825 // CHECK11-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 1826 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, i32* [[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]], i32* [[DOTCAPTURE_EXPR_1]], align 4 1831 // CHECK11-NEXT: store i32 0, i32* [[I]], align 4 1832 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, i32* [[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, i32* [[DOTOMP_LB]], align 4 1837 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1838 // CHECK11-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_UB]], align 4 1839 // CHECK11-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 1840 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 1841 // CHECK11-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_LB]], align 4 1842 // CHECK11-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_UB]], align 4 1843 // CHECK11-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1844 // CHECK11-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1845 // CHECK11-NEXT: [[TMP9:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 1846 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4 1847 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1848 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1849 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1850 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]] 1851 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1852 // CHECK11: cond.true: 1853 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 1854 // CHECK11-NEXT: br label [[COND_END:%.*]] 1855 // CHECK11: cond.false: 1856 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1857 // CHECK11-NEXT: br label [[COND_END]] 1858 // CHECK11: cond.end: 1859 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ] 1860 // CHECK11-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 1861 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 1862 // CHECK11-NEXT: store i32 [[TMP15]], i32* [[DOTOMP_IV]], align 4 1863 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1864 // CHECK11: omp.inner.for.cond: 1865 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1866 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 1867 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]] 1868 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1869 // CHECK11: omp.inner.for.body: 1870 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1871 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1 1872 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1873 // CHECK11-NEXT: store i32 [[ADD]], i32* [[I3]], align 4 1874 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, i32* [[I3]], align 4 1875 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i32 [[TMP19]] 1876 // CHECK11-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 1877 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1878 // CHECK11: omp.body.continue: 1879 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1880 // CHECK11: omp.inner.for.inc: 1881 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1882 // CHECK11-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP20]], 1 1883 // CHECK11-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4 1884 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]] 1885 // CHECK11: omp.inner.for.end: 1886 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1887 // CHECK11: omp.loop.exit: 1888 // CHECK11-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 1889 // CHECK11-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 1890 // CHECK11-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 1891 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 1892 // CHECK11: omp.precond.end: 1893 // CHECK11-NEXT: ret void 1894 // 1895 // 1896 // CHECK11-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 1897 // CHECK11-SAME: () #[[ATTR5:[0-9]+]] { 1898 // CHECK11-NEXT: entry: 1899 // CHECK11-NEXT: call void @__tgt_register_requires(i64 1) 1900 // CHECK11-NEXT: ret void 1901 // 1902 // 1903 // CHECK17-LABEL: define {{[^@]+}}@_Z21teams_template_structv 1904 // CHECK17-SAME: () #[[ATTR0:[0-9]+]] { 1905 // CHECK17-NEXT: entry: 1906 // CHECK17-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 1907 // CHECK17-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(%struct.SS* noundef nonnull align 4 dereferenceable(496) [[V]]) 1908 // CHECK17-NEXT: ret i32 [[CALL]] 1909 // 1910 // 1911 // CHECK17-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 1912 // CHECK17-SAME: (%struct.SS* noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 1913 // CHECK17-NEXT: entry: 1914 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 1915 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 8 1916 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 8 1917 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 8 1918 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4 1919 // CHECK17-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 1920 // CHECK17-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 1921 // CHECK17-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0 1922 // CHECK17-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1923 // CHECK17-NEXT: [[TMP1:%.*]] = bitcast i8** [[TMP0]] to %struct.SS** 1924 // CHECK17-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP1]], align 8 1925 // CHECK17-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1926 // CHECK17-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [123 x i32]** 1927 // CHECK17-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP3]], align 8 1928 // CHECK17-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 1929 // CHECK17-NEXT: store i8* null, i8** [[TMP4]], align 8 1930 // CHECK17-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1931 // CHECK17-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1932 // CHECK17-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 123) 1933 // CHECK17-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) 1934 // CHECK17-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 1935 // CHECK17-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1936 // CHECK17: omp_offload.failed: 1937 // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]] 1938 // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]] 1939 // CHECK17: omp_offload.cont: 1940 // CHECK17-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 1941 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i64 0, i64 0 1942 // CHECK17-NEXT: [[TMP9:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 1943 // CHECK17-NEXT: ret i32 [[TMP9]] 1944 // 1945 // 1946 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109 1947 // CHECK17-SAME: (%struct.SS* noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 1948 // CHECK17-NEXT: entry: 1949 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 1950 // CHECK17-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 1951 // CHECK17-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 1952 // CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.SS* [[TMP0]]) 1953 // CHECK17-NEXT: ret void 1954 // 1955 // 1956 // CHECK17-LABEL: define {{[^@]+}}@.omp_outlined. 1957 // CHECK17-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %struct.SS* noundef [[THIS:%.*]]) #[[ATTR1]] { 1958 // CHECK17-NEXT: entry: 1959 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 1960 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 1961 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 1962 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1963 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4 1964 // CHECK17-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 1965 // CHECK17-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 1966 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1967 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1968 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4 1969 // CHECK17-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 1970 // CHECK17-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 1971 // CHECK17-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 1972 // CHECK17-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 1973 // CHECK17-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 1974 // CHECK17-NEXT: store i32 122, i32* [[DOTOMP_COMB_UB]], align 4 1975 // CHECK17-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 1976 // CHECK17-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 1977 // CHECK17-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 1978 // CHECK17-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 1979 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 1980 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1981 // CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 1982 // CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1983 // CHECK17: cond.true: 1984 // CHECK17-NEXT: br label [[COND_END:%.*]] 1985 // CHECK17: cond.false: 1986 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1987 // CHECK17-NEXT: br label [[COND_END]] 1988 // CHECK17: cond.end: 1989 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 1990 // CHECK17-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 1991 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 1992 // CHECK17-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 1993 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1994 // CHECK17: omp.inner.for.cond: 1995 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 1996 // CHECK17-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 1997 // CHECK17-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 1998 // CHECK17-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1999 // CHECK17: omp.inner.for.body: 2000 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 2001 // CHECK17-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 2002 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2003 // CHECK17-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 2004 // CHECK17-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, %struct.SS*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], %struct.SS* [[TMP0]]) 2005 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2006 // CHECK17: omp.inner.for.inc: 2007 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2008 // CHECK17-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 2009 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] 2010 // CHECK17-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 2011 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]] 2012 // CHECK17: omp.inner.for.end: 2013 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2014 // CHECK17: omp.loop.exit: 2015 // CHECK17-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) 2016 // CHECK17-NEXT: ret void 2017 // 2018 // 2019 // CHECK17-LABEL: define {{[^@]+}}@.omp_outlined..1 2020 // CHECK17-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], %struct.SS* noundef [[THIS:%.*]]) #[[ATTR1]] { 2021 // CHECK17-NEXT: entry: 2022 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 2023 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 2024 // CHECK17-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 2025 // CHECK17-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 2026 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 8 2027 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2028 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4 2029 // CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2030 // CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2031 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2032 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2033 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4 2034 // CHECK17-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 2035 // CHECK17-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 2036 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 2037 // CHECK17-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 2038 // CHECK17-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 8 2039 // CHECK17-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 8 2040 // CHECK17-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 2041 // CHECK17-NEXT: store i32 122, i32* [[DOTOMP_UB]], align 4 2042 // CHECK17-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 2043 // CHECK17-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32 2044 // CHECK17-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 2045 // CHECK17-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32 2046 // CHECK17-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 2047 // CHECK17-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 2048 // CHECK17-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2049 // CHECK17-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2050 // CHECK17-NEXT: [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2051 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4 2052 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 2053 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2054 // CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122 2055 // CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2056 // CHECK17: cond.true: 2057 // CHECK17-NEXT: br label [[COND_END:%.*]] 2058 // CHECK17: cond.false: 2059 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2060 // CHECK17-NEXT: br label [[COND_END]] 2061 // CHECK17: cond.end: 2062 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 2063 // CHECK17-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 2064 // CHECK17-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2065 // CHECK17-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4 2066 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2067 // CHECK17: omp.inner.for.cond: 2068 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2069 // CHECK17-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2070 // CHECK17-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 2071 // CHECK17-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2072 // CHECK17: omp.inner.for.body: 2073 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2074 // CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 2075 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2076 // CHECK17-NEXT: store i32 [[ADD]], i32* [[I]], align 4 2077 // CHECK17-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0 2078 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4 2079 // CHECK17-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 2080 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A]], i64 0, i64 [[IDXPROM]] 2081 // CHECK17-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 2082 // CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2083 // CHECK17: omp.body.continue: 2084 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2085 // CHECK17: omp.inner.for.inc: 2086 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2087 // CHECK17-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1 2088 // CHECK17-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 2089 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]] 2090 // CHECK17: omp.inner.for.end: 2091 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2092 // CHECK17: omp.loop.exit: 2093 // CHECK17-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]]) 2094 // CHECK17-NEXT: ret void 2095 // 2096 // 2097 // CHECK17-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 2098 // CHECK17-SAME: () #[[ATTR3:[0-9]+]] { 2099 // CHECK17-NEXT: entry: 2100 // CHECK17-NEXT: call void @__tgt_register_requires(i64 1) 2101 // CHECK17-NEXT: ret void 2102 // 2103 // 2104 // CHECK19-LABEL: define {{[^@]+}}@_Z21teams_template_structv 2105 // CHECK19-SAME: () #[[ATTR0:[0-9]+]] { 2106 // CHECK19-NEXT: entry: 2107 // CHECK19-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 2108 // CHECK19-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(%struct.SS* noundef nonnull align 4 dereferenceable(496) [[V]]) 2109 // CHECK19-NEXT: ret i32 [[CALL]] 2110 // 2111 // 2112 // CHECK19-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 2113 // CHECK19-SAME: (%struct.SS* noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 2114 // CHECK19-NEXT: entry: 2115 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 2116 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [1 x i8*], align 4 2117 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [1 x i8*], align 4 2118 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [1 x i8*], align 4 2119 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4 2120 // CHECK19-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 2121 // CHECK19-NEXT: [[THIS1:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 2122 // CHECK19-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[THIS1]], i32 0, i32 0 2123 // CHECK19-NEXT: [[TMP0:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2124 // CHECK19-NEXT: [[TMP1:%.*]] = bitcast i8** [[TMP0]] to %struct.SS** 2125 // CHECK19-NEXT: store %struct.SS* [[THIS1]], %struct.SS** [[TMP1]], align 4 2126 // CHECK19-NEXT: [[TMP2:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2127 // CHECK19-NEXT: [[TMP3:%.*]] = bitcast i8** [[TMP2]] to [123 x i32]** 2128 // CHECK19-NEXT: store [123 x i32]* [[A]], [123 x i32]** [[TMP3]], align 4 2129 // CHECK19-NEXT: [[TMP4:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 2130 // CHECK19-NEXT: store i8* null, i8** [[TMP4]], align 4 2131 // CHECK19-NEXT: [[TMP5:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2132 // CHECK19-NEXT: [[TMP6:%.*]] = getelementptr inbounds [1 x i8*], [1 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2133 // CHECK19-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 123) 2134 // CHECK19-NEXT: [[TMP7:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109.region_id, i32 1, i8** [[TMP5]], i8** [[TMP6]], i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_sizes, i32 0, i32 0), i64* getelementptr inbounds ([1 x i64], [1 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) 2135 // CHECK19-NEXT: [[TMP8:%.*]] = icmp ne i32 [[TMP7]], 0 2136 // CHECK19-NEXT: br i1 [[TMP8]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2137 // CHECK19: omp_offload.failed: 2138 // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109(%struct.SS* [[THIS1]]) #[[ATTR2:[0-9]+]] 2139 // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]] 2140 // CHECK19: omp_offload.cont: 2141 // CHECK19-NEXT: [[A2:%.*]] = getelementptr inbounds [[STRUCT_SS]], %struct.SS* [[THIS1]], i32 0, i32 0 2142 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A2]], i32 0, i32 0 2143 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, i32* [[ARRAYIDX]], align 4 2144 // CHECK19-NEXT: ret i32 [[TMP9]] 2145 // 2146 // 2147 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l109 2148 // CHECK19-SAME: (%struct.SS* noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 2149 // CHECK19-NEXT: entry: 2150 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 2151 // CHECK19-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 2152 // CHECK19-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 2153 // CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, %struct.SS*)* @.omp_outlined. to void (i32*, i32*, ...)*), %struct.SS* [[TMP0]]) 2154 // CHECK19-NEXT: ret void 2155 // 2156 // 2157 // CHECK19-LABEL: define {{[^@]+}}@.omp_outlined. 2158 // CHECK19-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], %struct.SS* noundef [[THIS:%.*]]) #[[ATTR1]] { 2159 // CHECK19-NEXT: entry: 2160 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 2161 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 2162 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 2163 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2164 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4 2165 // CHECK19-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 2166 // CHECK19-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 2167 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2168 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2169 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4 2170 // CHECK19-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 2171 // CHECK19-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 2172 // CHECK19-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 2173 // CHECK19-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 2174 // CHECK19-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 2175 // CHECK19-NEXT: store i32 122, i32* [[DOTOMP_COMB_UB]], align 4 2176 // CHECK19-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2177 // CHECK19-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2178 // CHECK19-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 2179 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 2180 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 2181 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2182 // CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 2183 // CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2184 // CHECK19: cond.true: 2185 // CHECK19-NEXT: br label [[COND_END:%.*]] 2186 // CHECK19: cond.false: 2187 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2188 // CHECK19-NEXT: br label [[COND_END]] 2189 // CHECK19: cond.end: 2190 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2191 // CHECK19-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 2192 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 2193 // CHECK19-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 2194 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2195 // CHECK19: omp.inner.for.cond: 2196 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2197 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2198 // CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 2199 // CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2200 // CHECK19: omp.inner.for.body: 2201 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 2202 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2203 // CHECK19-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, %struct.SS*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], %struct.SS* [[TMP0]]) 2204 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2205 // CHECK19: omp.inner.for.inc: 2206 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2207 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 2208 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] 2209 // CHECK19-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 2210 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]] 2211 // CHECK19: omp.inner.for.end: 2212 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2213 // CHECK19: omp.loop.exit: 2214 // CHECK19-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) 2215 // CHECK19-NEXT: ret void 2216 // 2217 // 2218 // CHECK19-LABEL: define {{[^@]+}}@.omp_outlined..1 2219 // CHECK19-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], %struct.SS* noundef [[THIS:%.*]]) #[[ATTR1]] { 2220 // CHECK19-NEXT: entry: 2221 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 2222 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 2223 // CHECK19-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 2224 // CHECK19-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 2225 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca %struct.SS*, align 4 2226 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2227 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4 2228 // CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2229 // CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2230 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2231 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2232 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4 2233 // CHECK19-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 2234 // CHECK19-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 2235 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 2236 // CHECK19-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 2237 // CHECK19-NEXT: store %struct.SS* [[THIS]], %struct.SS** [[THIS_ADDR]], align 4 2238 // CHECK19-NEXT: [[TMP0:%.*]] = load %struct.SS*, %struct.SS** [[THIS_ADDR]], align 4 2239 // CHECK19-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 2240 // CHECK19-NEXT: store i32 122, i32* [[DOTOMP_UB]], align 4 2241 // CHECK19-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 2242 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 2243 // CHECK19-NEXT: store i32 [[TMP1]], i32* [[DOTOMP_LB]], align 4 2244 // CHECK19-NEXT: store i32 [[TMP2]], i32* [[DOTOMP_UB]], align 4 2245 // CHECK19-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2246 // CHECK19-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2247 // CHECK19-NEXT: [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 2248 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4 2249 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP4]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 2250 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2251 // CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 122 2252 // CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2253 // CHECK19: cond.true: 2254 // CHECK19-NEXT: br label [[COND_END:%.*]] 2255 // CHECK19: cond.false: 2256 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2257 // CHECK19-NEXT: br label [[COND_END]] 2258 // CHECK19: cond.end: 2259 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 2260 // CHECK19-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 2261 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2262 // CHECK19-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4 2263 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2264 // CHECK19: omp.inner.for.cond: 2265 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2266 // CHECK19-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2267 // CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 2268 // CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2269 // CHECK19: omp.inner.for.body: 2270 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2271 // CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 2272 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2273 // CHECK19-NEXT: store i32 [[ADD]], i32* [[I]], align 4 2274 // CHECK19-NEXT: [[A:%.*]] = getelementptr inbounds [[STRUCT_SS:%.*]], %struct.SS* [[TMP0]], i32 0, i32 0 2275 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4 2276 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], [123 x i32]* [[A]], i32 0, i32 [[TMP11]] 2277 // CHECK19-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 2278 // CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2279 // CHECK19: omp.body.continue: 2280 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2281 // CHECK19: omp.inner.for.inc: 2282 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2283 // CHECK19-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1 2284 // CHECK19-NEXT: store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4 2285 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]] 2286 // CHECK19: omp.inner.for.end: 2287 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2288 // CHECK19: omp.loop.exit: 2289 // CHECK19-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]]) 2290 // CHECK19-NEXT: ret void 2291 // 2292 // 2293 // CHECK19-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 2294 // CHECK19-SAME: () #[[ATTR3:[0-9]+]] { 2295 // CHECK19-NEXT: entry: 2296 // CHECK19-NEXT: call void @__tgt_register_requires(i64 1) 2297 // CHECK19-NEXT: ret void 2298 // 2299 // 2300 // CHECK25-LABEL: define {{[^@]+}}@main 2301 // CHECK25-SAME: (i32 noundef signext [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 2302 // CHECK25-NEXT: entry: 2303 // CHECK25-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 2304 // CHECK25-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 2305 // CHECK25-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 8 2306 // CHECK25-NEXT: [[N:%.*]] = alloca i32, align 4 2307 // CHECK25-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 8 2308 // CHECK25-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 2309 // CHECK25-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 2310 // CHECK25-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 2311 // CHECK25-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 2312 // CHECK25-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 2313 // CHECK25-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8 2314 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 2315 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2316 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2317 // CHECK25-NEXT: store i32 0, i32* [[RETVAL]], align 4 2318 // CHECK25-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 2319 // CHECK25-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 8 2320 // CHECK25-NEXT: store i32 100, i32* [[N]], align 4 2321 // CHECK25-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 2322 // CHECK25-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 2323 // CHECK25-NEXT: [[TMP2:%.*]] = call i8* @llvm.stacksave() 2324 // CHECK25-NEXT: store i8* [[TMP2]], i8** [[SAVED_STACK]], align 8 2325 // CHECK25-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 2326 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[__VLA_EXPR0]], align 8 2327 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, i32* [[N]], align 4 2328 // CHECK25-NEXT: [[CONV:%.*]] = bitcast i64* [[N_CASTED]] to i32* 2329 // CHECK25-NEXT: store i32 [[TMP3]], i32* [[CONV]], align 4 2330 // CHECK25-NEXT: [[TMP4:%.*]] = load i64, i64* [[N_CASTED]], align 8 2331 // CHECK25-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 2332 // CHECK25-NEXT: [[TMP6:%.*]] = bitcast [3 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 2333 // CHECK25-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 8 [[TMP6]], i8* align 8 bitcast ([3 x i64]* @.offload_sizes to i8*), i64 24, i1 false) 2334 // CHECK25-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2335 // CHECK25-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i64* 2336 // CHECK25-NEXT: store i64 [[TMP4]], i64* [[TMP8]], align 8 2337 // CHECK25-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2338 // CHECK25-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64* 2339 // CHECK25-NEXT: store i64 [[TMP4]], i64* [[TMP10]], align 8 2340 // CHECK25-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 2341 // CHECK25-NEXT: store i8* null, i8** [[TMP11]], align 8 2342 // CHECK25-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 2343 // CHECK25-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i64* 2344 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP13]], align 8 2345 // CHECK25-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 2346 // CHECK25-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i64* 2347 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP15]], align 8 2348 // CHECK25-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 2349 // CHECK25-NEXT: store i8* null, i8** [[TMP16]], align 8 2350 // CHECK25-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 2351 // CHECK25-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** 2352 // CHECK25-NEXT: store i32* [[VLA]], i32** [[TMP18]], align 8 2353 // CHECK25-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 2354 // CHECK25-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32** 2355 // CHECK25-NEXT: store i32* [[VLA]], i32** [[TMP20]], align 8 2356 // CHECK25-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 2357 // CHECK25-NEXT: store i64 [[TMP5]], i64* [[TMP21]], align 8 2358 // CHECK25-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 2359 // CHECK25-NEXT: store i8* null, i8** [[TMP22]], align 8 2360 // CHECK25-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2361 // CHECK25-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2362 // CHECK25-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 2363 // CHECK25-NEXT: [[TMP26:%.*]] = load i32, i32* [[N]], align 4 2364 // CHECK25-NEXT: store i32 [[TMP26]], i32* [[DOTCAPTURE_EXPR_]], align 4 2365 // CHECK25-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2366 // CHECK25-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP27]], 0 2367 // CHECK25-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2368 // CHECK25-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 2369 // CHECK25-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 2370 // CHECK25-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2371 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP28]], 1 2372 // CHECK25-NEXT: [[TMP29:%.*]] = zext i32 [[ADD]] to i64 2373 // CHECK25-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP29]]) 2374 // CHECK25-NEXT: [[TMP30:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP23]], i8** [[TMP24]], i64* [[TMP25]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) 2375 // CHECK25-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0 2376 // CHECK25-NEXT: br i1 [[TMP31]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2377 // CHECK25: omp_offload.failed: 2378 // CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i64 [[TMP4]], i64 [[TMP1]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] 2379 // CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]] 2380 // CHECK25: omp_offload.cont: 2381 // CHECK25-NEXT: [[TMP32:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 2382 // CHECK25-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10EEiT_(i32 noundef signext [[TMP32]]) 2383 // CHECK25-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 2384 // CHECK25-NEXT: [[TMP33:%.*]] = load i8*, i8** [[SAVED_STACK]], align 8 2385 // CHECK25-NEXT: call void @llvm.stackrestore(i8* [[TMP33]]) 2386 // CHECK25-NEXT: [[TMP34:%.*]] = load i32, i32* [[RETVAL]], align 4 2387 // CHECK25-NEXT: ret i32 [[TMP34]] 2388 // 2389 // 2390 // CHECK25-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162 2391 // CHECK25-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 2392 // CHECK25-NEXT: entry: 2393 // CHECK25-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 2394 // CHECK25-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 2395 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 2396 // CHECK25-NEXT: store i64 [[N]], i64* [[N_ADDR]], align 8 2397 // CHECK25-NEXT: store i64 [[VLA]], i64* [[VLA_ADDR]], align 8 2398 // CHECK25-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 2399 // CHECK25-NEXT: [[CONV:%.*]] = bitcast i64* [[N_ADDR]] to i32* 2400 // CHECK25-NEXT: [[TMP0:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 2401 // CHECK25-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 8 2402 // CHECK25-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i64, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[CONV]], i64 [[TMP0]], i32* [[TMP1]]) 2403 // CHECK25-NEXT: ret void 2404 // 2405 // 2406 // CHECK25-LABEL: define {{[^@]+}}@.omp_outlined. 2407 // CHECK25-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 2408 // CHECK25-NEXT: entry: 2409 // CHECK25-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 2410 // CHECK25-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 2411 // CHECK25-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 2412 // CHECK25-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 2413 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 2414 // CHECK25-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2415 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 2416 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2417 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2418 // CHECK25-NEXT: [[I:%.*]] = alloca i32, align 4 2419 // CHECK25-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 2420 // CHECK25-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 2421 // CHECK25-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2422 // CHECK25-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2423 // CHECK25-NEXT: [[I3:%.*]] = alloca i32, align 4 2424 // CHECK25-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 2425 // CHECK25-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 2426 // CHECK25-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 2427 // CHECK25-NEXT: store i64 [[VLA]], i64* [[VLA_ADDR]], align 8 2428 // CHECK25-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 2429 // CHECK25-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 2430 // CHECK25-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 2431 // CHECK25-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 8 2432 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 2433 // CHECK25-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 2434 // CHECK25-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2435 // CHECK25-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 2436 // CHECK25-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2437 // CHECK25-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 2438 // CHECK25-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 2439 // CHECK25-NEXT: store i32 0, i32* [[I]], align 4 2440 // CHECK25-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2441 // CHECK25-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 2442 // CHECK25-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2443 // CHECK25: omp.precond.then: 2444 // CHECK25-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 2445 // CHECK25-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2446 // CHECK25-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_COMB_UB]], align 4 2447 // CHECK25-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2448 // CHECK25-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2449 // CHECK25-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2450 // CHECK25-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 2451 // CHECK25-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 2452 // CHECK25-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2453 // CHECK25-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2454 // CHECK25-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 2455 // CHECK25-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2456 // CHECK25: cond.true: 2457 // CHECK25-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2458 // CHECK25-NEXT: br label [[COND_END:%.*]] 2459 // CHECK25: cond.false: 2460 // CHECK25-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2461 // CHECK25-NEXT: br label [[COND_END]] 2462 // CHECK25: cond.end: 2463 // CHECK25-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 2464 // CHECK25-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 2465 // CHECK25-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 2466 // CHECK25-NEXT: store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4 2467 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2468 // CHECK25: omp.inner.for.cond: 2469 // CHECK25-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2470 // CHECK25-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2471 // CHECK25-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 2472 // CHECK25-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2473 // CHECK25: omp.inner.for.body: 2474 // CHECK25-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 2475 // CHECK25-NEXT: [[TMP17:%.*]] = zext i32 [[TMP16]] to i64 2476 // CHECK25-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2477 // CHECK25-NEXT: [[TMP19:%.*]] = zext i32 [[TMP18]] to i64 2478 // CHECK25-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, i32*, i64, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i64 [[TMP17]], i64 [[TMP19]], i32* [[TMP0]], i64 [[TMP1]], i32* [[TMP2]]) 2479 // CHECK25-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2480 // CHECK25: omp.inner.for.inc: 2481 // CHECK25-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2482 // CHECK25-NEXT: [[TMP21:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 2483 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]] 2484 // CHECK25-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 2485 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND]] 2486 // CHECK25: omp.inner.for.end: 2487 // CHECK25-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2488 // CHECK25: omp.loop.exit: 2489 // CHECK25-NEXT: [[TMP22:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2490 // CHECK25-NEXT: [[TMP23:%.*]] = load i32, i32* [[TMP22]], align 4 2491 // CHECK25-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP23]]) 2492 // CHECK25-NEXT: br label [[OMP_PRECOND_END]] 2493 // CHECK25: omp.precond.end: 2494 // CHECK25-NEXT: ret void 2495 // 2496 // 2497 // CHECK25-LABEL: define {{[^@]+}}@.omp_outlined..1 2498 // CHECK25-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 2499 // CHECK25-NEXT: entry: 2500 // CHECK25-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 2501 // CHECK25-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 2502 // CHECK25-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 2503 // CHECK25-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 2504 // CHECK25-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 8 2505 // CHECK25-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 2506 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 8 2507 // CHECK25-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2508 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 2509 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2510 // CHECK25-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2511 // CHECK25-NEXT: [[I:%.*]] = alloca i32, align 4 2512 // CHECK25-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2513 // CHECK25-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2514 // CHECK25-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2515 // CHECK25-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2516 // CHECK25-NEXT: [[I4:%.*]] = alloca i32, align 4 2517 // CHECK25-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 2518 // CHECK25-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 2519 // CHECK25-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 2520 // CHECK25-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 2521 // CHECK25-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 8 2522 // CHECK25-NEXT: store i64 [[VLA]], i64* [[VLA_ADDR]], align 8 2523 // CHECK25-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 8 2524 // CHECK25-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 8 2525 // CHECK25-NEXT: [[TMP1:%.*]] = load i64, i64* [[VLA_ADDR]], align 8 2526 // CHECK25-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 8 2527 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 2528 // CHECK25-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 2529 // CHECK25-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2530 // CHECK25-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 2531 // CHECK25-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2532 // CHECK25-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 2533 // CHECK25-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 2534 // CHECK25-NEXT: store i32 0, i32* [[I]], align 4 2535 // CHECK25-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2536 // CHECK25-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 2537 // CHECK25-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2538 // CHECK25: omp.precond.then: 2539 // CHECK25-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 2540 // CHECK25-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2541 // CHECK25-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_UB]], align 4 2542 // CHECK25-NEXT: [[TMP7:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 2543 // CHECK25-NEXT: [[CONV:%.*]] = trunc i64 [[TMP7]] to i32 2544 // CHECK25-NEXT: [[TMP8:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 2545 // CHECK25-NEXT: [[CONV3:%.*]] = trunc i64 [[TMP8]] to i32 2546 // CHECK25-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 2547 // CHECK25-NEXT: store i32 [[CONV3]], i32* [[DOTOMP_UB]], align 4 2548 // CHECK25-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2549 // CHECK25-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2550 // CHECK25-NEXT: [[TMP9:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2551 // CHECK25-NEXT: [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4 2552 // CHECK25-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 2553 // CHECK25-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2554 // CHECK25-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2555 // CHECK25-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]] 2556 // CHECK25-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2557 // CHECK25: cond.true: 2558 // CHECK25-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2559 // CHECK25-NEXT: br label [[COND_END:%.*]] 2560 // CHECK25: cond.false: 2561 // CHECK25-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2562 // CHECK25-NEXT: br label [[COND_END]] 2563 // CHECK25: cond.end: 2564 // CHECK25-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ] 2565 // CHECK25-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 2566 // CHECK25-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2567 // CHECK25-NEXT: store i32 [[TMP15]], i32* [[DOTOMP_IV]], align 4 2568 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2569 // CHECK25: omp.inner.for.cond: 2570 // CHECK25-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2571 // CHECK25-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2572 // CHECK25-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]] 2573 // CHECK25-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2574 // CHECK25: omp.inner.for.body: 2575 // CHECK25-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2576 // CHECK25-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1 2577 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2578 // CHECK25-NEXT: store i32 [[ADD]], i32* [[I4]], align 4 2579 // CHECK25-NEXT: [[TMP19:%.*]] = load i32, i32* [[I4]], align 4 2580 // CHECK25-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP19]] to i64 2581 // CHECK25-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i64 [[IDXPROM]] 2582 // CHECK25-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 2583 // CHECK25-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2584 // CHECK25: omp.body.continue: 2585 // CHECK25-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2586 // CHECK25: omp.inner.for.inc: 2587 // CHECK25-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2588 // CHECK25-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP20]], 1 2589 // CHECK25-NEXT: store i32 [[ADD7]], i32* [[DOTOMP_IV]], align 4 2590 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND]] 2591 // CHECK25: omp.inner.for.end: 2592 // CHECK25-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2593 // CHECK25: omp.loop.exit: 2594 // CHECK25-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2595 // CHECK25-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 2596 // CHECK25-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 2597 // CHECK25-NEXT: br label [[OMP_PRECOND_END]] 2598 // CHECK25: omp.precond.end: 2599 // CHECK25-NEXT: ret void 2600 // 2601 // 2602 // CHECK25-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 2603 // CHECK25-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 2604 // CHECK25-NEXT: entry: 2605 // CHECK25-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 2606 // CHECK25-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 2607 // CHECK25-NEXT: [[TE:%.*]] = alloca i32, align 4 2608 // CHECK25-NEXT: [[TH:%.*]] = alloca i32, align 4 2609 // CHECK25-NEXT: [[TE_CASTED:%.*]] = alloca i64, align 8 2610 // CHECK25-NEXT: [[TH_CASTED:%.*]] = alloca i64, align 8 2611 // CHECK25-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 8 2612 // CHECK25-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 8 2613 // CHECK25-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 8 2614 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 2615 // CHECK25-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 2616 // CHECK25-NEXT: store i32 0, i32* [[TE]], align 4 2617 // CHECK25-NEXT: store i32 128, i32* [[TH]], align 4 2618 // CHECK25-NEXT: [[TMP0:%.*]] = load i32, i32* [[TE]], align 4 2619 // CHECK25-NEXT: [[CONV:%.*]] = bitcast i64* [[TE_CASTED]] to i32* 2620 // CHECK25-NEXT: store i32 [[TMP0]], i32* [[CONV]], align 4 2621 // CHECK25-NEXT: [[TMP1:%.*]] = load i64, i64* [[TE_CASTED]], align 8 2622 // CHECK25-NEXT: [[TMP2:%.*]] = load i32, i32* [[TH]], align 4 2623 // CHECK25-NEXT: [[CONV1:%.*]] = bitcast i64* [[TH_CASTED]] to i32* 2624 // CHECK25-NEXT: store i32 [[TMP2]], i32* [[CONV1]], align 4 2625 // CHECK25-NEXT: [[TMP3:%.*]] = load i64, i64* [[TH_CASTED]], align 8 2626 // CHECK25-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2627 // CHECK25-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i64* 2628 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP5]], align 8 2629 // CHECK25-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2630 // CHECK25-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i64* 2631 // CHECK25-NEXT: store i64 [[TMP1]], i64* [[TMP7]], align 8 2632 // CHECK25-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 2633 // CHECK25-NEXT: store i8* null, i8** [[TMP8]], align 8 2634 // CHECK25-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 2635 // CHECK25-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i64* 2636 // CHECK25-NEXT: store i64 [[TMP3]], i64* [[TMP10]], align 8 2637 // CHECK25-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 2638 // CHECK25-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i64* 2639 // CHECK25-NEXT: store i64 [[TMP3]], i64* [[TMP12]], align 8 2640 // CHECK25-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 2641 // CHECK25-NEXT: store i8* null, i8** [[TMP13]], align 8 2642 // CHECK25-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 2643 // CHECK25-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to [10 x i32]** 2644 // CHECK25-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP15]], align 8 2645 // CHECK25-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 2646 // CHECK25-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** 2647 // CHECK25-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 8 2648 // CHECK25-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 2649 // CHECK25-NEXT: store i8* null, i8** [[TMP18]], align 8 2650 // CHECK25-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2651 // CHECK25-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2652 // CHECK25-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 2653 // CHECK25-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) 2654 // CHECK25-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0) 2655 // CHECK25-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 2656 // CHECK25-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2657 // CHECK25: omp_offload.failed: 2658 // CHECK25-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i64 [[TMP1]], i64 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] 2659 // CHECK25-NEXT: br label [[OMP_OFFLOAD_CONT]] 2660 // CHECK25: omp_offload.cont: 2661 // CHECK25-NEXT: ret i32 0 2662 // 2663 // 2664 // CHECK25-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151 2665 // CHECK25-SAME: (i64 noundef [[TE:%.*]], i64 noundef [[TH:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2666 // CHECK25-NEXT: entry: 2667 // CHECK25-NEXT: [[TE_ADDR:%.*]] = alloca i64, align 8 2668 // CHECK25-NEXT: [[TH_ADDR:%.*]] = alloca i64, align 8 2669 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 2670 // CHECK25-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB3]]) 2671 // CHECK25-NEXT: store i64 [[TE]], i64* [[TE_ADDR]], align 8 2672 // CHECK25-NEXT: store i64 [[TH]], i64* [[TH_ADDR]], align 8 2673 // CHECK25-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 2674 // CHECK25-NEXT: [[CONV:%.*]] = bitcast i64* [[TE_ADDR]] to i32* 2675 // CHECK25-NEXT: [[CONV1:%.*]] = bitcast i64* [[TH_ADDR]] to i32* 2676 // CHECK25-NEXT: [[TMP1:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 2677 // CHECK25-NEXT: [[TMP2:%.*]] = load i32, i32* [[CONV]], align 4 2678 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, i32* [[CONV1]], align 4 2679 // CHECK25-NEXT: call void @__kmpc_push_num_teams(%struct.ident_t* @[[GLOB3]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 2680 // CHECK25-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP1]]) 2681 // CHECK25-NEXT: ret void 2682 // 2683 // 2684 // CHECK25-LABEL: define {{[^@]+}}@.omp_outlined..2 2685 // CHECK25-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2686 // CHECK25-NEXT: entry: 2687 // CHECK25-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 2688 // CHECK25-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 2689 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 2690 // CHECK25-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2691 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 2692 // CHECK25-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 2693 // CHECK25-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 2694 // CHECK25-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2695 // CHECK25-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2696 // CHECK25-NEXT: [[I:%.*]] = alloca i32, align 4 2697 // CHECK25-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 2698 // CHECK25-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 2699 // CHECK25-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 2700 // CHECK25-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 2701 // CHECK25-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 2702 // CHECK25-NEXT: store i32 9, i32* [[DOTOMP_COMB_UB]], align 4 2703 // CHECK25-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2704 // CHECK25-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2705 // CHECK25-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2706 // CHECK25-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 2707 // CHECK25-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 2708 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2709 // CHECK25-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 2710 // CHECK25-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2711 // CHECK25: cond.true: 2712 // CHECK25-NEXT: br label [[COND_END:%.*]] 2713 // CHECK25: cond.false: 2714 // CHECK25-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2715 // CHECK25-NEXT: br label [[COND_END]] 2716 // CHECK25: cond.end: 2717 // CHECK25-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2718 // CHECK25-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 2719 // CHECK25-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 2720 // CHECK25-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 2721 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2722 // CHECK25: omp.inner.for.cond: 2723 // CHECK25-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2724 // CHECK25-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2725 // CHECK25-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 2726 // CHECK25-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2727 // CHECK25: omp.inner.for.body: 2728 // CHECK25-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 2729 // CHECK25-NEXT: [[TMP9:%.*]] = zext i32 [[TMP8]] to i64 2730 // CHECK25-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2731 // CHECK25-NEXT: [[TMP11:%.*]] = zext i32 [[TMP10]] to i64 2732 // CHECK25-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i64, i64, [10 x i32]*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i64 [[TMP9]], i64 [[TMP11]], [10 x i32]* [[TMP0]]) 2733 // CHECK25-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2734 // CHECK25: omp.inner.for.inc: 2735 // CHECK25-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2736 // CHECK25-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 2737 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP12]], [[TMP13]] 2738 // CHECK25-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 2739 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND]] 2740 // CHECK25: omp.inner.for.end: 2741 // CHECK25-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2742 // CHECK25: omp.loop.exit: 2743 // CHECK25-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) 2744 // CHECK25-NEXT: ret void 2745 // 2746 // 2747 // CHECK25-LABEL: define {{[^@]+}}@.omp_outlined..3 2748 // CHECK25-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[DOTPREVIOUS_LB_:%.*]], i64 noundef [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 2749 // CHECK25-NEXT: entry: 2750 // CHECK25-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 8 2751 // CHECK25-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 8 2752 // CHECK25-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i64, align 8 2753 // CHECK25-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i64, align 8 2754 // CHECK25-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 8 2755 // CHECK25-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2756 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 2757 // CHECK25-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2758 // CHECK25-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2759 // CHECK25-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2760 // CHECK25-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2761 // CHECK25-NEXT: [[I:%.*]] = alloca i32, align 4 2762 // CHECK25-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 8 2763 // CHECK25-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 8 2764 // CHECK25-NEXT: store i64 [[DOTPREVIOUS_LB_]], i64* [[DOTPREVIOUS_LB__ADDR]], align 8 2765 // CHECK25-NEXT: store i64 [[DOTPREVIOUS_UB_]], i64* [[DOTPREVIOUS_UB__ADDR]], align 8 2766 // CHECK25-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 8 2767 // CHECK25-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 8 2768 // CHECK25-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 2769 // CHECK25-NEXT: store i32 9, i32* [[DOTOMP_UB]], align 4 2770 // CHECK25-NEXT: [[TMP1:%.*]] = load i64, i64* [[DOTPREVIOUS_LB__ADDR]], align 8 2771 // CHECK25-NEXT: [[CONV:%.*]] = trunc i64 [[TMP1]] to i32 2772 // CHECK25-NEXT: [[TMP2:%.*]] = load i64, i64* [[DOTPREVIOUS_UB__ADDR]], align 8 2773 // CHECK25-NEXT: [[CONV1:%.*]] = trunc i64 [[TMP2]] to i32 2774 // CHECK25-NEXT: store i32 [[CONV]], i32* [[DOTOMP_LB]], align 4 2775 // CHECK25-NEXT: store i32 [[CONV1]], i32* [[DOTOMP_UB]], align 4 2776 // CHECK25-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2777 // CHECK25-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2778 // CHECK25-NEXT: [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 8 2779 // CHECK25-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4 2780 // CHECK25-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP4]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 2781 // CHECK25-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2782 // CHECK25-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9 2783 // CHECK25-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2784 // CHECK25: cond.true: 2785 // CHECK25-NEXT: br label [[COND_END:%.*]] 2786 // CHECK25: cond.false: 2787 // CHECK25-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2788 // CHECK25-NEXT: br label [[COND_END]] 2789 // CHECK25: cond.end: 2790 // CHECK25-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 2791 // CHECK25-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 2792 // CHECK25-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 2793 // CHECK25-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4 2794 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2795 // CHECK25: omp.inner.for.cond: 2796 // CHECK25-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2797 // CHECK25-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 2798 // CHECK25-NEXT: [[CMP2:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 2799 // CHECK25-NEXT: br i1 [[CMP2]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2800 // CHECK25: omp.inner.for.body: 2801 // CHECK25-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2802 // CHECK25-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 2803 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2804 // CHECK25-NEXT: store i32 [[ADD]], i32* [[I]], align 4 2805 // CHECK25-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4 2806 // CHECK25-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 2807 // CHECK25-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], [10 x i32]* [[TMP0]], i64 0, i64 [[IDXPROM]] 2808 // CHECK25-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 2809 // CHECK25-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2810 // CHECK25: omp.body.continue: 2811 // CHECK25-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2812 // CHECK25: omp.inner.for.inc: 2813 // CHECK25-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2814 // CHECK25-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP12]], 1 2815 // CHECK25-NEXT: store i32 [[ADD3]], i32* [[DOTOMP_IV]], align 4 2816 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND]] 2817 // CHECK25: omp.inner.for.end: 2818 // CHECK25-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2819 // CHECK25: omp.loop.exit: 2820 // CHECK25-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]]) 2821 // CHECK25-NEXT: ret void 2822 // 2823 // 2824 // CHECK25-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 2825 // CHECK25-SAME: () #[[ATTR6:[0-9]+]] { 2826 // CHECK25-NEXT: entry: 2827 // CHECK25-NEXT: call void @__tgt_register_requires(i64 1) 2828 // CHECK25-NEXT: ret void 2829 // 2830 // 2831 // CHECK27-LABEL: define {{[^@]+}}@main 2832 // CHECK27-SAME: (i32 noundef [[ARGC:%.*]], i8** noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 2833 // CHECK27-NEXT: entry: 2834 // CHECK27-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 2835 // CHECK27-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 2836 // CHECK27-NEXT: [[ARGV_ADDR:%.*]] = alloca i8**, align 4 2837 // CHECK27-NEXT: [[N:%.*]] = alloca i32, align 4 2838 // CHECK27-NEXT: [[SAVED_STACK:%.*]] = alloca i8*, align 4 2839 // CHECK27-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 2840 // CHECK27-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 2841 // CHECK27-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 2842 // CHECK27-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 2843 // CHECK27-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 2844 // CHECK27-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4 2845 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 2846 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2847 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2848 // CHECK27-NEXT: store i32 0, i32* [[RETVAL]], align 4 2849 // CHECK27-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 2850 // CHECK27-NEXT: store i8** [[ARGV]], i8*** [[ARGV_ADDR]], align 4 2851 // CHECK27-NEXT: store i32 100, i32* [[N]], align 4 2852 // CHECK27-NEXT: [[TMP0:%.*]] = load i32, i32* [[N]], align 4 2853 // CHECK27-NEXT: [[TMP1:%.*]] = call i8* @llvm.stacksave() 2854 // CHECK27-NEXT: store i8* [[TMP1]], i8** [[SAVED_STACK]], align 4 2855 // CHECK27-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 2856 // CHECK27-NEXT: store i32 [[TMP0]], i32* [[__VLA_EXPR0]], align 4 2857 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, i32* [[N]], align 4 2858 // CHECK27-NEXT: store i32 [[TMP2]], i32* [[N_CASTED]], align 4 2859 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, i32* [[N_CASTED]], align 4 2860 // CHECK27-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 2861 // CHECK27-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 2862 // CHECK27-NEXT: [[TMP6:%.*]] = bitcast [3 x i64]* [[DOTOFFLOAD_SIZES]] to i8* 2863 // CHECK27-NEXT: call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 [[TMP6]], i8* align 4 bitcast ([3 x i64]* @.offload_sizes to i8*), i32 24, i1 false) 2864 // CHECK27-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2865 // CHECK27-NEXT: [[TMP8:%.*]] = bitcast i8** [[TMP7]] to i32* 2866 // CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP8]], align 4 2867 // CHECK27-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2868 // CHECK27-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32* 2869 // CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP10]], align 4 2870 // CHECK27-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 2871 // CHECK27-NEXT: store i8* null, i8** [[TMP11]], align 4 2872 // CHECK27-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 2873 // CHECK27-NEXT: [[TMP13:%.*]] = bitcast i8** [[TMP12]] to i32* 2874 // CHECK27-NEXT: store i32 [[TMP0]], i32* [[TMP13]], align 4 2875 // CHECK27-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 2876 // CHECK27-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to i32* 2877 // CHECK27-NEXT: store i32 [[TMP0]], i32* [[TMP15]], align 4 2878 // CHECK27-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 2879 // CHECK27-NEXT: store i8* null, i8** [[TMP16]], align 4 2880 // CHECK27-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 2881 // CHECK27-NEXT: [[TMP18:%.*]] = bitcast i8** [[TMP17]] to i32** 2882 // CHECK27-NEXT: store i32* [[VLA]], i32** [[TMP18]], align 4 2883 // CHECK27-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 2884 // CHECK27-NEXT: [[TMP20:%.*]] = bitcast i8** [[TMP19]] to i32** 2885 // CHECK27-NEXT: store i32* [[VLA]], i32** [[TMP20]], align 4 2886 // CHECK27-NEXT: [[TMP21:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 2 2887 // CHECK27-NEXT: store i64 [[TMP5]], i64* [[TMP21]], align 4 2888 // CHECK27-NEXT: [[TMP22:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 2889 // CHECK27-NEXT: store i8* null, i8** [[TMP22]], align 4 2890 // CHECK27-NEXT: [[TMP23:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2891 // CHECK27-NEXT: [[TMP24:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2892 // CHECK27-NEXT: [[TMP25:%.*]] = getelementptr inbounds [3 x i64], [3 x i64]* [[DOTOFFLOAD_SIZES]], i32 0, i32 0 2893 // CHECK27-NEXT: [[TMP26:%.*]] = load i32, i32* [[N]], align 4 2894 // CHECK27-NEXT: store i32 [[TMP26]], i32* [[DOTCAPTURE_EXPR_]], align 4 2895 // CHECK27-NEXT: [[TMP27:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2896 // CHECK27-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP27]], 0 2897 // CHECK27-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2898 // CHECK27-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 2899 // CHECK27-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 2900 // CHECK27-NEXT: [[TMP28:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2901 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP28]], 1 2902 // CHECK27-NEXT: [[TMP29:%.*]] = zext i32 [[ADD]] to i64 2903 // CHECK27-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3:[0-9]+]], i64 -1, i64 [[TMP29]]) 2904 // CHECK27-NEXT: [[TMP30:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162.region_id, i32 3, i8** [[TMP23]], i8** [[TMP24]], i64* [[TMP25]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes, i32 0, i32 0), i8** null, i8** null, i32 0, i32 0) 2905 // CHECK27-NEXT: [[TMP31:%.*]] = icmp ne i32 [[TMP30]], 0 2906 // CHECK27-NEXT: br i1 [[TMP31]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2907 // CHECK27: omp_offload.failed: 2908 // CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162(i32 [[TMP3]], i32 [[TMP0]], i32* [[VLA]]) #[[ATTR3:[0-9]+]] 2909 // CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]] 2910 // CHECK27: omp_offload.cont: 2911 // CHECK27-NEXT: [[TMP32:%.*]] = load i32, i32* [[ARGC_ADDR]], align 4 2912 // CHECK27-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10EEiT_(i32 noundef [[TMP32]]) 2913 // CHECK27-NEXT: store i32 [[CALL]], i32* [[RETVAL]], align 4 2914 // CHECK27-NEXT: [[TMP33:%.*]] = load i8*, i8** [[SAVED_STACK]], align 4 2915 // CHECK27-NEXT: call void @llvm.stackrestore(i8* [[TMP33]]) 2916 // CHECK27-NEXT: [[TMP34:%.*]] = load i32, i32* [[RETVAL]], align 4 2917 // CHECK27-NEXT: ret i32 [[TMP34]] 2918 // 2919 // 2920 // CHECK27-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l162 2921 // CHECK27-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 2922 // CHECK27-NEXT: entry: 2923 // CHECK27-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 2924 // CHECK27-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 2925 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 2926 // CHECK27-NEXT: store i32 [[N]], i32* [[N_ADDR]], align 4 2927 // CHECK27-NEXT: store i32 [[VLA]], i32* [[VLA_ADDR]], align 4 2928 // CHECK27-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 2929 // CHECK27-NEXT: [[TMP0:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 2930 // CHECK27-NEXT: [[TMP1:%.*]] = load i32*, i32** [[A_ADDR]], align 4 2931 // CHECK27-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32*, i32, i32*)* @.omp_outlined. to void (i32*, i32*, ...)*), i32* [[N_ADDR]], i32 [[TMP0]], i32* [[TMP1]]) 2932 // CHECK27-NEXT: ret void 2933 // 2934 // 2935 // CHECK27-LABEL: define {{[^@]+}}@.omp_outlined. 2936 // CHECK27-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 2937 // CHECK27-NEXT: entry: 2938 // CHECK27-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 2939 // CHECK27-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 2940 // CHECK27-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 2941 // CHECK27-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 2942 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 2943 // CHECK27-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2944 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 2945 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 2946 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 2947 // CHECK27-NEXT: [[I:%.*]] = alloca i32, align 4 2948 // CHECK27-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 2949 // CHECK27-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 2950 // CHECK27-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2951 // CHECK27-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2952 // CHECK27-NEXT: [[I3:%.*]] = alloca i32, align 4 2953 // CHECK27-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 2954 // CHECK27-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 2955 // CHECK27-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 2956 // CHECK27-NEXT: store i32 [[VLA]], i32* [[VLA_ADDR]], align 4 2957 // CHECK27-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 2958 // CHECK27-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 2959 // CHECK27-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 2960 // CHECK27-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 4 2961 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 2962 // CHECK27-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 2963 // CHECK27-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2964 // CHECK27-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 2965 // CHECK27-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 2966 // CHECK27-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 2967 // CHECK27-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 2968 // CHECK27-NEXT: store i32 0, i32* [[I]], align 4 2969 // CHECK27-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 2970 // CHECK27-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 2971 // CHECK27-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 2972 // CHECK27: omp.precond.then: 2973 // CHECK27-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 2974 // CHECK27-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2975 // CHECK27-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_COMB_UB]], align 4 2976 // CHECK27-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 2977 // CHECK27-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 2978 // CHECK27-NEXT: [[TMP7:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 2979 // CHECK27-NEXT: [[TMP8:%.*]] = load i32, i32* [[TMP7]], align 4 2980 // CHECK27-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 2981 // CHECK27-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2982 // CHECK27-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2983 // CHECK27-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 2984 // CHECK27-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2985 // CHECK27: cond.true: 2986 // CHECK27-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 2987 // CHECK27-NEXT: br label [[COND_END:%.*]] 2988 // CHECK27: cond.false: 2989 // CHECK27-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 2990 // CHECK27-NEXT: br label [[COND_END]] 2991 // CHECK27: cond.end: 2992 // CHECK27-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 2993 // CHECK27-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 2994 // CHECK27-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 2995 // CHECK27-NEXT: store i32 [[TMP13]], i32* [[DOTOMP_IV]], align 4 2996 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2997 // CHECK27: omp.inner.for.cond: 2998 // CHECK27-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 2999 // CHECK27-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 3000 // CHECK27-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 3001 // CHECK27-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3002 // CHECK27: omp.inner.for.body: 3003 // CHECK27-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 3004 // CHECK27-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 3005 // CHECK27-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 5, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, i32*, i32, i32*)* @.omp_outlined..1 to void (i32*, i32*, ...)*), i32 [[TMP16]], i32 [[TMP17]], i32* [[TMP0]], i32 [[TMP1]], i32* [[TMP2]]) 3006 // CHECK27-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3007 // CHECK27: omp.inner.for.inc: 3008 // CHECK27-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 3009 // CHECK27-NEXT: [[TMP19:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 3010 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP18]], [[TMP19]] 3011 // CHECK27-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 3012 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND]] 3013 // CHECK27: omp.inner.for.end: 3014 // CHECK27-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3015 // CHECK27: omp.loop.exit: 3016 // CHECK27-NEXT: [[TMP20:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3017 // CHECK27-NEXT: [[TMP21:%.*]] = load i32, i32* [[TMP20]], align 4 3018 // CHECK27-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP21]]) 3019 // CHECK27-NEXT: br label [[OMP_PRECOND_END]] 3020 // CHECK27: omp.precond.end: 3021 // CHECK27-NEXT: ret void 3022 // 3023 // 3024 // CHECK27-LABEL: define {{[^@]+}}@.omp_outlined..1 3025 // CHECK27-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], i32* noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 3026 // CHECK27-NEXT: entry: 3027 // CHECK27-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 3028 // CHECK27-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 3029 // CHECK27-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 3030 // CHECK27-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 3031 // CHECK27-NEXT: [[N_ADDR:%.*]] = alloca i32*, align 4 3032 // CHECK27-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 3033 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca i32*, align 4 3034 // CHECK27-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3035 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 3036 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3037 // CHECK27-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3038 // CHECK27-NEXT: [[I:%.*]] = alloca i32, align 4 3039 // CHECK27-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3040 // CHECK27-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3041 // CHECK27-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3042 // CHECK27-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3043 // CHECK27-NEXT: [[I3:%.*]] = alloca i32, align 4 3044 // CHECK27-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 3045 // CHECK27-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 3046 // CHECK27-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 3047 // CHECK27-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 3048 // CHECK27-NEXT: store i32* [[N]], i32** [[N_ADDR]], align 4 3049 // CHECK27-NEXT: store i32 [[VLA]], i32* [[VLA_ADDR]], align 4 3050 // CHECK27-NEXT: store i32* [[A]], i32** [[A_ADDR]], align 4 3051 // CHECK27-NEXT: [[TMP0:%.*]] = load i32*, i32** [[N_ADDR]], align 4 3052 // CHECK27-NEXT: [[TMP1:%.*]] = load i32, i32* [[VLA_ADDR]], align 4 3053 // CHECK27-NEXT: [[TMP2:%.*]] = load i32*, i32** [[A_ADDR]], align 4 3054 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, i32* [[TMP0]], align 4 3055 // CHECK27-NEXT: store i32 [[TMP3]], i32* [[DOTCAPTURE_EXPR_]], align 4 3056 // CHECK27-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3057 // CHECK27-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 3058 // CHECK27-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3059 // CHECK27-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3060 // CHECK27-NEXT: store i32 [[SUB2]], i32* [[DOTCAPTURE_EXPR_1]], align 4 3061 // CHECK27-NEXT: store i32 0, i32* [[I]], align 4 3062 // CHECK27-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_]], align 4 3063 // CHECK27-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 3064 // CHECK27-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3065 // CHECK27: omp.precond.then: 3066 // CHECK27-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 3067 // CHECK27-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3068 // CHECK27-NEXT: store i32 [[TMP6]], i32* [[DOTOMP_UB]], align 4 3069 // CHECK27-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 3070 // CHECK27-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 3071 // CHECK27-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_LB]], align 4 3072 // CHECK27-NEXT: store i32 [[TMP8]], i32* [[DOTOMP_UB]], align 4 3073 // CHECK27-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 3074 // CHECK27-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 3075 // CHECK27-NEXT: [[TMP9:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3076 // CHECK27-NEXT: [[TMP10:%.*]] = load i32, i32* [[TMP9]], align 4 3077 // CHECK27-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2:[0-9]+]], i32 [[TMP10]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 3078 // CHECK27-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3079 // CHECK27-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3080 // CHECK27-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP11]], [[TMP12]] 3081 // CHECK27-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3082 // CHECK27: cond.true: 3083 // CHECK27-NEXT: [[TMP13:%.*]] = load i32, i32* [[DOTCAPTURE_EXPR_1]], align 4 3084 // CHECK27-NEXT: br label [[COND_END:%.*]] 3085 // CHECK27: cond.false: 3086 // CHECK27-NEXT: [[TMP14:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3087 // CHECK27-NEXT: br label [[COND_END]] 3088 // CHECK27: cond.end: 3089 // CHECK27-NEXT: [[COND:%.*]] = phi i32 [ [[TMP13]], [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ] 3090 // CHECK27-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 3091 // CHECK27-NEXT: [[TMP15:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 3092 // CHECK27-NEXT: store i32 [[TMP15]], i32* [[DOTOMP_IV]], align 4 3093 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3094 // CHECK27: omp.inner.for.cond: 3095 // CHECK27-NEXT: [[TMP16:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 3096 // CHECK27-NEXT: [[TMP17:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3097 // CHECK27-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]] 3098 // CHECK27-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3099 // CHECK27: omp.inner.for.body: 3100 // CHECK27-NEXT: [[TMP18:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 3101 // CHECK27-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1 3102 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3103 // CHECK27-NEXT: store i32 [[ADD]], i32* [[I3]], align 4 3104 // CHECK27-NEXT: [[TMP19:%.*]] = load i32, i32* [[I3]], align 4 3105 // CHECK27-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, i32* [[TMP2]], i32 [[TMP19]] 3106 // CHECK27-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 3107 // CHECK27-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3108 // CHECK27: omp.body.continue: 3109 // CHECK27-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3110 // CHECK27: omp.inner.for.inc: 3111 // CHECK27-NEXT: [[TMP20:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 3112 // CHECK27-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP20]], 1 3113 // CHECK27-NEXT: store i32 [[ADD6]], i32* [[DOTOMP_IV]], align 4 3114 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND]] 3115 // CHECK27: omp.inner.for.end: 3116 // CHECK27-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3117 // CHECK27: omp.loop.exit: 3118 // CHECK27-NEXT: [[TMP21:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3119 // CHECK27-NEXT: [[TMP22:%.*]] = load i32, i32* [[TMP21]], align 4 3120 // CHECK27-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP22]]) 3121 // CHECK27-NEXT: br label [[OMP_PRECOND_END]] 3122 // CHECK27: omp.precond.end: 3123 // CHECK27-NEXT: ret void 3124 // 3125 // 3126 // CHECK27-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 3127 // CHECK27-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 3128 // CHECK27-NEXT: entry: 3129 // CHECK27-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 3130 // CHECK27-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 3131 // CHECK27-NEXT: [[TE:%.*]] = alloca i32, align 4 3132 // CHECK27-NEXT: [[TH:%.*]] = alloca i32, align 4 3133 // CHECK27-NEXT: [[TE_CASTED:%.*]] = alloca i32, align 4 3134 // CHECK27-NEXT: [[TH_CASTED:%.*]] = alloca i32, align 4 3135 // CHECK27-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x i8*], align 4 3136 // CHECK27-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x i8*], align 4 3137 // CHECK27-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x i8*], align 4 3138 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 3139 // CHECK27-NEXT: store i32 [[ARGC]], i32* [[ARGC_ADDR]], align 4 3140 // CHECK27-NEXT: store i32 0, i32* [[TE]], align 4 3141 // CHECK27-NEXT: store i32 128, i32* [[TH]], align 4 3142 // CHECK27-NEXT: [[TMP0:%.*]] = load i32, i32* [[TE]], align 4 3143 // CHECK27-NEXT: store i32 [[TMP0]], i32* [[TE_CASTED]], align 4 3144 // CHECK27-NEXT: [[TMP1:%.*]] = load i32, i32* [[TE_CASTED]], align 4 3145 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, i32* [[TH]], align 4 3146 // CHECK27-NEXT: store i32 [[TMP2]], i32* [[TH_CASTED]], align 4 3147 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, i32* [[TH_CASTED]], align 4 3148 // CHECK27-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3149 // CHECK27-NEXT: [[TMP5:%.*]] = bitcast i8** [[TMP4]] to i32* 3150 // CHECK27-NEXT: store i32 [[TMP1]], i32* [[TMP5]], align 4 3151 // CHECK27-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3152 // CHECK27-NEXT: [[TMP7:%.*]] = bitcast i8** [[TMP6]] to i32* 3153 // CHECK27-NEXT: store i32 [[TMP1]], i32* [[TMP7]], align 4 3154 // CHECK27-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 3155 // CHECK27-NEXT: store i8* null, i8** [[TMP8]], align 4 3156 // CHECK27-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 3157 // CHECK27-NEXT: [[TMP10:%.*]] = bitcast i8** [[TMP9]] to i32* 3158 // CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP10]], align 4 3159 // CHECK27-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 1 3160 // CHECK27-NEXT: [[TMP12:%.*]] = bitcast i8** [[TMP11]] to i32* 3161 // CHECK27-NEXT: store i32 [[TMP3]], i32* [[TMP12]], align 4 3162 // CHECK27-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 3163 // CHECK27-NEXT: store i8* null, i8** [[TMP13]], align 4 3164 // CHECK27-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 3165 // CHECK27-NEXT: [[TMP15:%.*]] = bitcast i8** [[TMP14]] to [10 x i32]** 3166 // CHECK27-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP15]], align 4 3167 // CHECK27-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 2 3168 // CHECK27-NEXT: [[TMP17:%.*]] = bitcast i8** [[TMP16]] to [10 x i32]** 3169 // CHECK27-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[TMP17]], align 4 3170 // CHECK27-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 3171 // CHECK27-NEXT: store i8* null, i8** [[TMP18]], align 4 3172 // CHECK27-NEXT: [[TMP19:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3173 // CHECK27-NEXT: [[TMP20:%.*]] = getelementptr inbounds [3 x i8*], [3 x i8*]* [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3174 // CHECK27-NEXT: [[TMP21:%.*]] = load i32, i32* [[TE]], align 4 3175 // CHECK27-NEXT: call void @__kmpc_push_target_tripcount_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i64 10) 3176 // CHECK27-NEXT: [[TMP22:%.*]] = call i32 @__tgt_target_teams_mapper(%struct.ident_t* @[[GLOB3]], i64 -1, i8* @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151.region_id, i32 3, i8** [[TMP19]], i8** [[TMP20]], i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_sizes.4, i32 0, i32 0), i64* getelementptr inbounds ([3 x i64], [3 x i64]* @.offload_maptypes.5, i32 0, i32 0), i8** null, i8** null, i32 [[TMP21]], i32 0) 3177 // CHECK27-NEXT: [[TMP23:%.*]] = icmp ne i32 [[TMP22]], 0 3178 // CHECK27-NEXT: br i1 [[TMP23]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 3179 // CHECK27: omp_offload.failed: 3180 // CHECK27-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151(i32 [[TMP1]], i32 [[TMP3]], [10 x i32]* [[A]]) #[[ATTR3]] 3181 // CHECK27-NEXT: br label [[OMP_OFFLOAD_CONT]] 3182 // CHECK27: omp_offload.cont: 3183 // CHECK27-NEXT: ret i32 0 3184 // 3185 // 3186 // CHECK27-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l151 3187 // CHECK27-SAME: (i32 noundef [[TE:%.*]], i32 noundef [[TH:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 3188 // CHECK27-NEXT: entry: 3189 // CHECK27-NEXT: [[TE_ADDR:%.*]] = alloca i32, align 4 3190 // CHECK27-NEXT: [[TH_ADDR:%.*]] = alloca i32, align 4 3191 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 3192 // CHECK27-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB3]]) 3193 // CHECK27-NEXT: store i32 [[TE]], i32* [[TE_ADDR]], align 4 3194 // CHECK27-NEXT: store i32 [[TH]], i32* [[TH_ADDR]], align 4 3195 // CHECK27-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 3196 // CHECK27-NEXT: [[TMP1:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 3197 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, i32* [[TE_ADDR]], align 4 3198 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, i32* [[TH_ADDR]], align 4 3199 // CHECK27-NEXT: call void @__kmpc_push_num_teams(%struct.ident_t* @[[GLOB3]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 3200 // CHECK27-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_teams(%struct.ident_t* @[[GLOB3]], i32 1, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, [10 x i32]*)* @.omp_outlined..2 to void (i32*, i32*, ...)*), [10 x i32]* [[TMP1]]) 3201 // CHECK27-NEXT: ret void 3202 // 3203 // 3204 // CHECK27-LABEL: define {{[^@]+}}@.omp_outlined..2 3205 // CHECK27-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 3206 // CHECK27-NEXT: entry: 3207 // CHECK27-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 3208 // CHECK27-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 3209 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 3210 // CHECK27-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3211 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 3212 // CHECK27-NEXT: [[DOTOMP_COMB_LB:%.*]] = alloca i32, align 4 3213 // CHECK27-NEXT: [[DOTOMP_COMB_UB:%.*]] = alloca i32, align 4 3214 // CHECK27-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3215 // CHECK27-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3216 // CHECK27-NEXT: [[I:%.*]] = alloca i32, align 4 3217 // CHECK27-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 3218 // CHECK27-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 3219 // CHECK27-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 3220 // CHECK27-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 3221 // CHECK27-NEXT: store i32 0, i32* [[DOTOMP_COMB_LB]], align 4 3222 // CHECK27-NEXT: store i32 9, i32* [[DOTOMP_COMB_UB]], align 4 3223 // CHECK27-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 3224 // CHECK27-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 3225 // CHECK27-NEXT: [[TMP1:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3226 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, i32* [[TMP1]], align 4 3227 // CHECK27-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]], i32 92, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_COMB_LB]], i32* [[DOTOMP_COMB_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 3228 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 3229 // CHECK27-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 3230 // CHECK27-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3231 // CHECK27: cond.true: 3232 // CHECK27-NEXT: br label [[COND_END:%.*]] 3233 // CHECK27: cond.false: 3234 // CHECK27-NEXT: [[TMP4:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 3235 // CHECK27-NEXT: br label [[COND_END]] 3236 // CHECK27: cond.end: 3237 // CHECK27-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 3238 // CHECK27-NEXT: store i32 [[COND]], i32* [[DOTOMP_COMB_UB]], align 4 3239 // CHECK27-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 3240 // CHECK27-NEXT: store i32 [[TMP5]], i32* [[DOTOMP_IV]], align 4 3241 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3242 // CHECK27: omp.inner.for.cond: 3243 // CHECK27-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 3244 // CHECK27-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 3245 // CHECK27-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 3246 // CHECK27-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3247 // CHECK27: omp.inner.for.body: 3248 // CHECK27-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_COMB_LB]], align 4 3249 // CHECK27-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_COMB_UB]], align 4 3250 // CHECK27-NEXT: call void (%struct.ident_t*, i32, void (i32*, i32*, ...)*, ...) @__kmpc_fork_call(%struct.ident_t* @[[GLOB3]], i32 3, void (i32*, i32*, ...)* bitcast (void (i32*, i32*, i32, i32, [10 x i32]*)* @.omp_outlined..3 to void (i32*, i32*, ...)*), i32 [[TMP8]], i32 [[TMP9]], [10 x i32]* [[TMP0]]) 3251 // CHECK27-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3252 // CHECK27: omp.inner.for.inc: 3253 // CHECK27-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 3254 // CHECK27-NEXT: [[TMP11:%.*]] = load i32, i32* [[DOTOMP_STRIDE]], align 4 3255 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP10]], [[TMP11]] 3256 // CHECK27-NEXT: store i32 [[ADD]], i32* [[DOTOMP_IV]], align 4 3257 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND]] 3258 // CHECK27: omp.inner.for.end: 3259 // CHECK27-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3260 // CHECK27: omp.loop.exit: 3261 // CHECK27-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP2]]) 3262 // CHECK27-NEXT: ret void 3263 // 3264 // 3265 // CHECK27-LABEL: define {{[^@]+}}@.omp_outlined..3 3266 // CHECK27-SAME: (i32* noalias noundef [[DOTGLOBAL_TID_:%.*]], i32* noalias noundef [[DOTBOUND_TID_:%.*]], i32 noundef [[DOTPREVIOUS_LB_:%.*]], i32 noundef [[DOTPREVIOUS_UB_:%.*]], [10 x i32]* noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 3267 // CHECK27-NEXT: entry: 3268 // CHECK27-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca i32*, align 4 3269 // CHECK27-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca i32*, align 4 3270 // CHECK27-NEXT: [[DOTPREVIOUS_LB__ADDR:%.*]] = alloca i32, align 4 3271 // CHECK27-NEXT: [[DOTPREVIOUS_UB__ADDR:%.*]] = alloca i32, align 4 3272 // CHECK27-NEXT: [[A_ADDR:%.*]] = alloca [10 x i32]*, align 4 3273 // CHECK27-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3274 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 3275 // CHECK27-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3276 // CHECK27-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3277 // CHECK27-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3278 // CHECK27-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3279 // CHECK27-NEXT: [[I:%.*]] = alloca i32, align 4 3280 // CHECK27-NEXT: store i32* [[DOTGLOBAL_TID_]], i32** [[DOTGLOBAL_TID__ADDR]], align 4 3281 // CHECK27-NEXT: store i32* [[DOTBOUND_TID_]], i32** [[DOTBOUND_TID__ADDR]], align 4 3282 // CHECK27-NEXT: store i32 [[DOTPREVIOUS_LB_]], i32* [[DOTPREVIOUS_LB__ADDR]], align 4 3283 // CHECK27-NEXT: store i32 [[DOTPREVIOUS_UB_]], i32* [[DOTPREVIOUS_UB__ADDR]], align 4 3284 // CHECK27-NEXT: store [10 x i32]* [[A]], [10 x i32]** [[A_ADDR]], align 4 3285 // CHECK27-NEXT: [[TMP0:%.*]] = load [10 x i32]*, [10 x i32]** [[A_ADDR]], align 4 3286 // CHECK27-NEXT: store i32 0, i32* [[DOTOMP_LB]], align 4 3287 // CHECK27-NEXT: store i32 9, i32* [[DOTOMP_UB]], align 4 3288 // CHECK27-NEXT: [[TMP1:%.*]] = load i32, i32* [[DOTPREVIOUS_LB__ADDR]], align 4 3289 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, i32* [[DOTPREVIOUS_UB__ADDR]], align 4 3290 // CHECK27-NEXT: store i32 [[TMP1]], i32* [[DOTOMP_LB]], align 4 3291 // CHECK27-NEXT: store i32 [[TMP2]], i32* [[DOTOMP_UB]], align 4 3292 // CHECK27-NEXT: store i32 1, i32* [[DOTOMP_STRIDE]], align 4 3293 // CHECK27-NEXT: store i32 0, i32* [[DOTOMP_IS_LAST]], align 4 3294 // CHECK27-NEXT: [[TMP3:%.*]] = load i32*, i32** [[DOTGLOBAL_TID__ADDR]], align 4 3295 // CHECK27-NEXT: [[TMP4:%.*]] = load i32, i32* [[TMP3]], align 4 3296 // CHECK27-NEXT: call void @__kmpc_for_static_init_4(%struct.ident_t* @[[GLOB2]], i32 [[TMP4]], i32 34, i32* [[DOTOMP_IS_LAST]], i32* [[DOTOMP_LB]], i32* [[DOTOMP_UB]], i32* [[DOTOMP_STRIDE]], i32 1, i32 1) 3297 // CHECK27-NEXT: [[TMP5:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3298 // CHECK27-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP5]], 9 3299 // CHECK27-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3300 // CHECK27: cond.true: 3301 // CHECK27-NEXT: br label [[COND_END:%.*]] 3302 // CHECK27: cond.false: 3303 // CHECK27-NEXT: [[TMP6:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3304 // CHECK27-NEXT: br label [[COND_END]] 3305 // CHECK27: cond.end: 3306 // CHECK27-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP6]], [[COND_FALSE]] ] 3307 // CHECK27-NEXT: store i32 [[COND]], i32* [[DOTOMP_UB]], align 4 3308 // CHECK27-NEXT: [[TMP7:%.*]] = load i32, i32* [[DOTOMP_LB]], align 4 3309 // CHECK27-NEXT: store i32 [[TMP7]], i32* [[DOTOMP_IV]], align 4 3310 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3311 // CHECK27: omp.inner.for.cond: 3312 // CHECK27-NEXT: [[TMP8:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 3313 // CHECK27-NEXT: [[TMP9:%.*]] = load i32, i32* [[DOTOMP_UB]], align 4 3314 // CHECK27-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 3315 // CHECK27-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3316 // CHECK27: omp.inner.for.body: 3317 // CHECK27-NEXT: [[TMP10:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 3318 // CHECK27-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 3319 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3320 // CHECK27-NEXT: store i32 [[ADD]], i32* [[I]], align 4 3321 // CHECK27-NEXT: [[TMP11:%.*]] = load i32, i32* [[I]], align 4 3322 // CHECK27-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], [10 x i32]* [[TMP0]], i32 0, i32 [[TMP11]] 3323 // CHECK27-NEXT: store i32 0, i32* [[ARRAYIDX]], align 4 3324 // CHECK27-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3325 // CHECK27: omp.body.continue: 3326 // CHECK27-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3327 // CHECK27: omp.inner.for.inc: 3328 // CHECK27-NEXT: [[TMP12:%.*]] = load i32, i32* [[DOTOMP_IV]], align 4 3329 // CHECK27-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1 3330 // CHECK27-NEXT: store i32 [[ADD2]], i32* [[DOTOMP_IV]], align 4 3331 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND]] 3332 // CHECK27: omp.inner.for.end: 3333 // CHECK27-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3334 // CHECK27: omp.loop.exit: 3335 // CHECK27-NEXT: call void @__kmpc_for_static_fini(%struct.ident_t* @[[GLOB1]], i32 [[TMP4]]) 3336 // CHECK27-NEXT: ret void 3337 // 3338 // 3339 // CHECK27-LABEL: define {{[^@]+}}@.omp_offloading.requires_reg 3340 // CHECK27-SAME: () #[[ATTR6:[0-9]+]] { 3341 // CHECK27-NEXT: entry: 3342 // CHECK27-NEXT: call void @__tgt_register_requires(i64 1) 3343 // CHECK27-NEXT: ret void 3344 // 3345