1 // NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _ 2 // expected-no-diagnostics 3 #ifndef HEADER 4 #define HEADER 5 // Test host codegen. 6 // RUN: %clang_cc1 -DCK1 -verify -Wno-vla -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK1 7 // RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 8 // RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK1 9 // RUN: %clang_cc1 -DCK1 -verify -Wno-vla -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK3 10 // RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 11 // RUN: %clang_cc1 -DCK1 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK3 12 13 // RUN: %clang_cc1 -DCK1 -verify -Wno-vla -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK5 14 // RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 15 // RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK5 16 // RUN: %clang_cc1 -DCK1 -verify -Wno-vla -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK7 17 // RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 18 // RUN: %clang_cc1 -DCK1 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK7 19 #ifdef CK1 20 21 int a[100]; 22 23 int teams_argument_global(int n) { 24 int i; 25 int te = n / 128; 26 int th = 128; 27 // discard n_addr and i 28 29 30 #pragma omp target 31 #pragma omp teams distribute simd num_teams(te), thread_limit(th) aligned(a) simdlen(16) linear(i) 32 for(i = 0; i < n; i++) { 33 a[i] = 0; 34 } 35 36 #pragma omp target 37 {{{ 38 #pragma omp teams distribute simd safelen(32) 39 for(int i = 0; i < n; i++) { 40 a[i] = 0; 41 } 42 }}} 43 44 // outlined target regions 45 46 47 48 49 return a[0]; 50 } 51 52 53 #endif // CK1 54 55 // Test host codegen. 56 // RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK9 57 // RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 58 // RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK9 59 // RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK11 60 // RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 61 // RUN: %clang_cc1 -DCK2 -fopenmp -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK11 62 63 // RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK13 64 // RUN: %clang_cc1 -DCK2 -fopenmp-simd -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 65 // RUN: %clang_cc1 -DCK2 -fopenmp-simd -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK13 66 // RUN: %clang_cc1 -DCK2 -verify -Wno-vla -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK15 67 // RUN: %clang_cc1 -DCK2 -fopenmp-simd -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 68 // RUN: %clang_cc1 -DCK2 -fopenmp-simd -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK15 69 #ifdef CK2 70 71 int teams_local_arg(void) { 72 int n = 100; 73 int a[n]; 74 75 #pragma omp target 76 #pragma omp teams distribute simd 77 for(int i = 0; i < n; i++) { 78 a[i] = 0; 79 } 80 81 // outlined target region 82 83 84 return a[0]; 85 } 86 #endif // CK2 87 88 // Test host codegen. 89 // RUN: %clang_cc1 -DCK3 -verify -Wno-vla -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK17 90 // RUN: %clang_cc1 -DCK3 -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 91 // RUN: %clang_cc1 -DCK3 -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK17 92 // RUN: %clang_cc1 -DCK3 -verify -Wno-vla -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK19 93 // RUN: %clang_cc1 -DCK3 -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 94 // RUN: %clang_cc1 -DCK3 -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK19 95 // RUN: %clang_cc1 -DCK3 -verify -Wno-vla -fopenmp -DOMP5 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK21 96 // RUN: %clang_cc1 -DCK3 -fopenmp -DOMP5 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 97 // RUN: %clang_cc1 -DCK3 -fopenmp -DOMP5 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK21 98 // RUN: %clang_cc1 -DCK3 -verify -Wno-vla -fopenmp -DOMP5 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK23 99 // RUN: %clang_cc1 -DCK3 -fopenmp -DOMP5 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 100 // RUN: %clang_cc1 -DCK3 -fopenmp -DOMP5 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK23 101 102 // RUN: %clang_cc1 -DCK3 -verify -Wno-vla -fopenmp-simd -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK25 103 // RUN: %clang_cc1 -DCK3 -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 104 // RUN: %clang_cc1 -DCK3 -fopenmp-simd -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK25 105 // RUN: %clang_cc1 -DCK3 -verify -Wno-vla -fopenmp-simd -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK27 106 // RUN: %clang_cc1 -DCK3 -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 107 // RUN: %clang_cc1 -DCK3 -fopenmp-simd -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK27 108 // RUN: %clang_cc1 -DCK3 -verify -Wno-vla -fopenmp-simd -DOMP5 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK29 109 // RUN: %clang_cc1 -DCK3 -fopenmp-simd -DOMP5 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 110 // RUN: %clang_cc1 -DCK3 -fopenmp-simd -DOMP5 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK29 111 // RUN: %clang_cc1 -DCK3 -verify -Wno-vla -fopenmp-simd -DOMP5 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK31 112 // RUN: %clang_cc1 -DCK3 -fopenmp-simd -DOMP5 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 113 // RUN: %clang_cc1 -DCK3 -fopenmp-simd -DOMP5 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK31 114 #ifdef CK3 115 116 117 template <typename T, int X, long long Y> 118 struct SS{ 119 T a[X]; 120 float b; 121 int foo(void) { 122 123 #pragma omp target 124 #ifdef OMP5 125 #pragma omp teams distribute simd if(b) nontemporal(a, b) 126 #else 127 #pragma omp teams distribute simd 128 #endif // OMP5 129 for(int i = 0; i < X; i++) { 130 a[i] = (T)b; 131 } 132 133 // outlined target region 134 135 136 return a[0]; 137 } 138 }; 139 140 int teams_template_struct(void) { 141 SS<int, 123, 456> V; 142 return V.foo(); 143 144 } 145 #endif // CK3 146 147 // Test host codegen. 148 // RUN: %clang_cc1 -DCK4 -verify -Wno-vla -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK33 149 // RUN: %clang_cc1 -DCK4 -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 150 // RUN: %clang_cc1 -DCK4 -fopenmp -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK33 151 // RUN: %clang_cc1 -DCK4 -verify -Wno-vla -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK35 152 // RUN: %clang_cc1 -DCK4 -fopenmp -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 153 // RUN: %clang_cc1 -DCK4 -fopenmp -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK35 154 // RUN: %clang_cc1 -DCK4 -verify -Wno-vla -fopenmp -DOMP5 -x c++ -triple powerpc64le-unknown-unknown -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK37 155 // RUN: %clang_cc1 -DCK4 -fopenmp -DOMP5 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 156 // RUN: %clang_cc1 -DCK4 -fopenmp -DOMP5 -x c++ -triple powerpc64le-unknown-unknown -fopenmp -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK37 157 // RUN: %clang_cc1 -DCK4 -verify -Wno-vla -fopenmp -DOMP5 -x c++ -triple i386-unknown-unknown -fopenmp -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK39 158 // RUN: %clang_cc1 -DCK4 -fopenmp -DOMP5 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 159 // RUN: %clang_cc1 -DCK4 -fopenmp -DOMP5 -x c++ -triple i386-unknown-unknown -fopenmp -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK39 160 161 // RUN: %clang_cc1 -DCK4 -verify -Wno-vla -fopenmp-simd -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK41 162 // RUN: %clang_cc1 -DCK4 -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 163 // RUN: %clang_cc1 -DCK4 -fopenmp-simd -fopenmp-version=45 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK41 164 // RUN: %clang_cc1 -DCK4 -verify -Wno-vla -fopenmp-simd -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK43 165 // RUN: %clang_cc1 -DCK4 -fopenmp-simd -fopenmp-version=45 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 166 // RUN: %clang_cc1 -DCK4 -fopenmp-simd -fopenmp-version=45 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK43 167 // RUN: %clang_cc1 -DCK4 -verify -Wno-vla -fopenmp-simd -DOMP5 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK45 168 // RUN: %clang_cc1 -DCK4 -fopenmp-simd -DOMP5 -x c++ -std=c++11 -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -emit-pch -o %t %s 169 // RUN: %clang_cc1 -DCK4 -fopenmp-simd -DOMP5 -x c++ -triple powerpc64le-unknown-unknown -fopenmp-targets=powerpc64le-ibm-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK45 170 // RUN: %clang_cc1 -DCK4 -verify -Wno-vla -fopenmp-simd -DOMP5 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-llvm %s -o - | FileCheck %s --check-prefix=CHECK47 171 // RUN: %clang_cc1 -DCK4 -fopenmp-simd -DOMP5 -x c++ -std=c++11 -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -emit-pch -o %t %s 172 // RUN: %clang_cc1 -DCK4 -fopenmp-simd -DOMP5 -x c++ -triple i386-unknown-unknown -fopenmp-targets=i386-pc-linux-gnu -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --check-prefix=CHECK47 173 174 #ifdef CK4 175 176 template <typename T, int n> 177 int tmain(T argc) { 178 T a[n]; 179 int te = n/128; 180 int th = 128; 181 #pragma omp target 182 #pragma omp teams distribute simd num_teams(te) thread_limit(th) 183 for(int i = 0; i < n; i++) { 184 a[i] = (T)0; 185 } 186 return 0; 187 } 188 189 int main (int argc, char **argv) { 190 int n = 100; 191 int a[n]; 192 #pragma omp target 193 #ifdef OMP5 194 #pragma omp teams distribute simd if(simd:argc) 195 #else 196 #pragma omp teams distribute simd 197 #endif // OMP5 198 for(int i = 0; i < n; i++) { 199 a[i] = 0; 200 } 201 return tmain<int, 10>(argc); 202 } 203 204 205 206 207 208 209 210 // OMP5_50-DAG: !{!"llvm.loop.vectorize.enable", i1 false} 211 #endif // CK4 212 #endif 213 // CHECK1-LABEL: define {{[^@]+}}@_Z21teams_argument_globali 214 // CHECK1-SAME: (i32 noundef signext [[N:%.*]]) #[[ATTR0:[0-9]+]] { 215 // CHECK1-NEXT: entry: 216 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 217 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 218 // CHECK1-NEXT: [[TE:%.*]] = alloca i32, align 4 219 // CHECK1-NEXT: [[TH:%.*]] = alloca i32, align 4 220 // CHECK1-NEXT: [[TE_CASTED:%.*]] = alloca i64, align 8 221 // CHECK1-NEXT: [[TH_CASTED:%.*]] = alloca i64, align 8 222 // CHECK1-NEXT: [[I_CASTED:%.*]] = alloca i64, align 8 223 // CHECK1-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 224 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x ptr], align 8 225 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x ptr], align 8 226 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x ptr], align 8 227 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 228 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 229 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 230 // CHECK1-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 231 // CHECK1-NEXT: [[N_CASTED4:%.*]] = alloca i64, align 8 232 // CHECK1-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [2 x ptr], align 8 233 // CHECK1-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [2 x ptr], align 8 234 // CHECK1-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [2 x ptr], align 8 235 // CHECK1-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 236 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 237 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 238 // CHECK1-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 239 // CHECK1-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 240 // CHECK1-NEXT: [[TMP0:%.*]] = load i32, ptr [[N_ADDR]], align 4 241 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP0]], 128 242 // CHECK1-NEXT: store i32 [[DIV]], ptr [[TE]], align 4 243 // CHECK1-NEXT: store i32 128, ptr [[TH]], align 4 244 // CHECK1-NEXT: [[TMP1:%.*]] = load i32, ptr [[TE]], align 4 245 // CHECK1-NEXT: store i32 [[TMP1]], ptr [[TE_CASTED]], align 4 246 // CHECK1-NEXT: [[TMP2:%.*]] = load i64, ptr [[TE_CASTED]], align 8 247 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[TH]], align 4 248 // CHECK1-NEXT: store i32 [[TMP3]], ptr [[TH_CASTED]], align 4 249 // CHECK1-NEXT: [[TMP4:%.*]] = load i64, ptr [[TH_CASTED]], align 8 250 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[I]], align 4 251 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[I_CASTED]], align 4 252 // CHECK1-NEXT: [[TMP6:%.*]] = load i64, ptr [[I_CASTED]], align 8 253 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[N_ADDR]], align 4 254 // CHECK1-NEXT: store i32 [[TMP7]], ptr [[N_CASTED]], align 4 255 // CHECK1-NEXT: [[TMP8:%.*]] = load i64, ptr [[N_CASTED]], align 8 256 // CHECK1-NEXT: [[TMP9:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 257 // CHECK1-NEXT: store i64 [[TMP2]], ptr [[TMP9]], align 8 258 // CHECK1-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 259 // CHECK1-NEXT: store i64 [[TMP2]], ptr [[TMP10]], align 8 260 // CHECK1-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 261 // CHECK1-NEXT: store ptr null, ptr [[TMP11]], align 8 262 // CHECK1-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 263 // CHECK1-NEXT: store i64 [[TMP4]], ptr [[TMP12]], align 8 264 // CHECK1-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 265 // CHECK1-NEXT: store i64 [[TMP4]], ptr [[TMP13]], align 8 266 // CHECK1-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 267 // CHECK1-NEXT: store ptr null, ptr [[TMP14]], align 8 268 // CHECK1-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 269 // CHECK1-NEXT: store ptr @a, ptr [[TMP15]], align 8 270 // CHECK1-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 271 // CHECK1-NEXT: store ptr @a, ptr [[TMP16]], align 8 272 // CHECK1-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 273 // CHECK1-NEXT: store ptr null, ptr [[TMP17]], align 8 274 // CHECK1-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 275 // CHECK1-NEXT: store i64 [[TMP6]], ptr [[TMP18]], align 8 276 // CHECK1-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3 277 // CHECK1-NEXT: store i64 [[TMP6]], ptr [[TMP19]], align 8 278 // CHECK1-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 279 // CHECK1-NEXT: store ptr null, ptr [[TMP20]], align 8 280 // CHECK1-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 281 // CHECK1-NEXT: store i64 [[TMP8]], ptr [[TMP21]], align 8 282 // CHECK1-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4 283 // CHECK1-NEXT: store i64 [[TMP8]], ptr [[TMP22]], align 8 284 // CHECK1-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 4 285 // CHECK1-NEXT: store ptr null, ptr [[TMP23]], align 8 286 // CHECK1-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 287 // CHECK1-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 288 // CHECK1-NEXT: [[TMP26:%.*]] = load i32, ptr [[TE]], align 4 289 // CHECK1-NEXT: [[TMP27:%.*]] = load i32, ptr [[TH]], align 4 290 // CHECK1-NEXT: [[TMP28:%.*]] = icmp ult i32 [[TMP27]], 1 291 // CHECK1-NEXT: [[TMP29:%.*]] = select i1 [[TMP28]], i32 [[TMP27]], i32 1 292 // CHECK1-NEXT: [[TMP30:%.*]] = load i32, ptr [[N_ADDR]], align 4 293 // CHECK1-NEXT: store i32 [[TMP30]], ptr [[DOTCAPTURE_EXPR_]], align 4 294 // CHECK1-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 295 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP31]], 0 296 // CHECK1-NEXT: [[DIV2:%.*]] = sdiv i32 [[SUB]], 1 297 // CHECK1-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV2]], 1 298 // CHECK1-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_1]], align 4 299 // CHECK1-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 300 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP32]], 1 301 // CHECK1-NEXT: [[TMP33:%.*]] = zext i32 [[ADD]] to i64 302 // CHECK1-NEXT: [[TMP34:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP26]], 0 303 // CHECK1-NEXT: [[TMP35:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP29]], 0 304 // CHECK1-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 305 // CHECK1-NEXT: store i32 3, ptr [[TMP36]], align 4 306 // CHECK1-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 307 // CHECK1-NEXT: store i32 5, ptr [[TMP37]], align 4 308 // CHECK1-NEXT: [[TMP38:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 309 // CHECK1-NEXT: store ptr [[TMP24]], ptr [[TMP38]], align 8 310 // CHECK1-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 311 // CHECK1-NEXT: store ptr [[TMP25]], ptr [[TMP39]], align 8 312 // CHECK1-NEXT: [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 313 // CHECK1-NEXT: store ptr @.offload_sizes, ptr [[TMP40]], align 8 314 // CHECK1-NEXT: [[TMP41:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 315 // CHECK1-NEXT: store ptr @.offload_maptypes, ptr [[TMP41]], align 8 316 // CHECK1-NEXT: [[TMP42:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 317 // CHECK1-NEXT: store ptr null, ptr [[TMP42]], align 8 318 // CHECK1-NEXT: [[TMP43:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 319 // CHECK1-NEXT: store ptr null, ptr [[TMP43]], align 8 320 // CHECK1-NEXT: [[TMP44:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 321 // CHECK1-NEXT: store i64 [[TMP33]], ptr [[TMP44]], align 8 322 // CHECK1-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 323 // CHECK1-NEXT: store i64 0, ptr [[TMP45]], align 8 324 // CHECK1-NEXT: [[TMP46:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 325 // CHECK1-NEXT: store [3 x i32] [[TMP34]], ptr [[TMP46]], align 4 326 // CHECK1-NEXT: [[TMP47:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 327 // CHECK1-NEXT: store [3 x i32] [[TMP35]], ptr [[TMP47]], align 4 328 // CHECK1-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 329 // CHECK1-NEXT: store i32 0, ptr [[TMP48]], align 4 330 // CHECK1-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 [[TMP26]], i32 [[TMP29]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l30.region_id, ptr [[KERNEL_ARGS]]) 331 // CHECK1-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 332 // CHECK1-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 333 // CHECK1: omp_offload.failed: 334 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l30(i64 [[TMP2]], i64 [[TMP4]], ptr @a, i64 [[TMP6]], i64 [[TMP8]]) #[[ATTR3:[0-9]+]] 335 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT]] 336 // CHECK1: omp_offload.cont: 337 // CHECK1-NEXT: [[TMP51:%.*]] = load i32, ptr [[N_ADDR]], align 4 338 // CHECK1-NEXT: store i32 [[TMP51]], ptr [[N_CASTED4]], align 4 339 // CHECK1-NEXT: [[TMP52:%.*]] = load i64, ptr [[N_CASTED4]], align 8 340 // CHECK1-NEXT: [[TMP53:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 341 // CHECK1-NEXT: store i64 [[TMP52]], ptr [[TMP53]], align 8 342 // CHECK1-NEXT: [[TMP54:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 343 // CHECK1-NEXT: store i64 [[TMP52]], ptr [[TMP54]], align 8 344 // CHECK1-NEXT: [[TMP55:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 0 345 // CHECK1-NEXT: store ptr null, ptr [[TMP55]], align 8 346 // CHECK1-NEXT: [[TMP56:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 347 // CHECK1-NEXT: store ptr @a, ptr [[TMP56]], align 8 348 // CHECK1-NEXT: [[TMP57:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 349 // CHECK1-NEXT: store ptr @a, ptr [[TMP57]], align 8 350 // CHECK1-NEXT: [[TMP58:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS7]], i64 0, i64 1 351 // CHECK1-NEXT: store ptr null, ptr [[TMP58]], align 8 352 // CHECK1-NEXT: [[TMP59:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 353 // CHECK1-NEXT: [[TMP60:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 354 // CHECK1-NEXT: [[TMP61:%.*]] = load i32, ptr [[N_ADDR]], align 4 355 // CHECK1-NEXT: store i32 [[TMP61]], ptr [[DOTCAPTURE_EXPR_9]], align 4 356 // CHECK1-NEXT: [[TMP62:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_9]], align 4 357 // CHECK1-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP62]], 0 358 // CHECK1-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 359 // CHECK1-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 360 // CHECK1-NEXT: store i32 [[SUB13]], ptr [[DOTCAPTURE_EXPR_10]], align 4 361 // CHECK1-NEXT: [[TMP63:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_10]], align 4 362 // CHECK1-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP63]], 1 363 // CHECK1-NEXT: [[TMP64:%.*]] = zext i32 [[ADD14]] to i64 364 // CHECK1-NEXT: [[TMP65:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0 365 // CHECK1-NEXT: store i32 3, ptr [[TMP65]], align 4 366 // CHECK1-NEXT: [[TMP66:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1 367 // CHECK1-NEXT: store i32 2, ptr [[TMP66]], align 4 368 // CHECK1-NEXT: [[TMP67:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2 369 // CHECK1-NEXT: store ptr [[TMP59]], ptr [[TMP67]], align 8 370 // CHECK1-NEXT: [[TMP68:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3 371 // CHECK1-NEXT: store ptr [[TMP60]], ptr [[TMP68]], align 8 372 // CHECK1-NEXT: [[TMP69:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4 373 // CHECK1-NEXT: store ptr @.offload_sizes.1, ptr [[TMP69]], align 8 374 // CHECK1-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5 375 // CHECK1-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP70]], align 8 376 // CHECK1-NEXT: [[TMP71:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6 377 // CHECK1-NEXT: store ptr null, ptr [[TMP71]], align 8 378 // CHECK1-NEXT: [[TMP72:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7 379 // CHECK1-NEXT: store ptr null, ptr [[TMP72]], align 8 380 // CHECK1-NEXT: [[TMP73:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8 381 // CHECK1-NEXT: store i64 [[TMP64]], ptr [[TMP73]], align 8 382 // CHECK1-NEXT: [[TMP74:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9 383 // CHECK1-NEXT: store i64 0, ptr [[TMP74]], align 8 384 // CHECK1-NEXT: [[TMP75:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10 385 // CHECK1-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP75]], align 4 386 // CHECK1-NEXT: [[TMP76:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11 387 // CHECK1-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP76]], align 4 388 // CHECK1-NEXT: [[TMP77:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12 389 // CHECK1-NEXT: store i32 0, ptr [[TMP77]], align 4 390 // CHECK1-NEXT: [[TMP78:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l36.region_id, ptr [[KERNEL_ARGS15]]) 391 // CHECK1-NEXT: [[TMP79:%.*]] = icmp ne i32 [[TMP78]], 0 392 // CHECK1-NEXT: br i1 [[TMP79]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] 393 // CHECK1: omp_offload.failed16: 394 // CHECK1-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l36(i64 [[TMP52]], ptr @a) #[[ATTR3]] 395 // CHECK1-NEXT: br label [[OMP_OFFLOAD_CONT17]] 396 // CHECK1: omp_offload.cont17: 397 // CHECK1-NEXT: [[TMP80:%.*]] = load i32, ptr @a, align 4 398 // CHECK1-NEXT: ret i32 [[TMP80]] 399 // 400 // 401 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l30 402 // CHECK1-SAME: (i64 noundef [[TE:%.*]], i64 noundef [[TH:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]], i64 noundef [[I:%.*]], i64 noundef [[N:%.*]]) #[[ATTR1:[0-9]+]] { 403 // CHECK1-NEXT: entry: 404 // CHECK1-NEXT: [[TE_ADDR:%.*]] = alloca i64, align 8 405 // CHECK1-NEXT: [[TH_ADDR:%.*]] = alloca i64, align 8 406 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 407 // CHECK1-NEXT: [[I_ADDR:%.*]] = alloca i64, align 8 408 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 409 // CHECK1-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2]]) 410 // CHECK1-NEXT: store i64 [[TE]], ptr [[TE_ADDR]], align 8 411 // CHECK1-NEXT: store i64 [[TH]], ptr [[TH_ADDR]], align 8 412 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 413 // CHECK1-NEXT: store i64 [[I]], ptr [[I_ADDR]], align 8 414 // CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 415 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 416 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TE_ADDR]], align 4 417 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[TH_ADDR]], align 4 418 // CHECK1-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB2]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 419 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l30.omp_outlined, ptr [[I_ADDR]], ptr [[N_ADDR]], ptr [[TMP1]]) 420 // CHECK1-NEXT: ret void 421 // 422 // 423 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l30.omp_outlined 424 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[I:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 425 // CHECK1-NEXT: entry: 426 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 427 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 428 // CHECK1-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 8 429 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8 430 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 431 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 432 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 433 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 434 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 435 // CHECK1-NEXT: [[I3:%.*]] = alloca i32, align 4 436 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 437 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 438 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 439 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 440 // CHECK1-NEXT: [[I4:%.*]] = alloca i32, align 4 441 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 442 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 443 // CHECK1-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 8 444 // CHECK1-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8 445 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 446 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 8 447 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[N_ADDR]], align 8 448 // CHECK1-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 449 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1]], align 4 450 // CHECK1-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 451 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 452 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 453 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 454 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 455 // CHECK1-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 456 // CHECK1-NEXT: store i32 0, ptr [[I3]], align 4 457 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 458 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 459 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 460 // CHECK1: omp.precond.then: 461 // CHECK1-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [100 x i32], ptr [[TMP2]], i64 0, i64 0 462 // CHECK1-NEXT: call void @llvm.assume(i1 true) [ "align"(ptr [[ARRAYDECAY]], i64 16) ] 463 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 464 // CHECK1-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 465 // CHECK1-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 466 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 467 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 468 // CHECK1-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 469 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 470 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 471 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 472 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 473 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 474 // CHECK1-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 475 // CHECK1: cond.true: 476 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 477 // CHECK1-NEXT: br label [[COND_END:%.*]] 478 // CHECK1: cond.false: 479 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 480 // CHECK1-NEXT: br label [[COND_END]] 481 // CHECK1: cond.end: 482 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 483 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 484 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 485 // CHECK1-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 486 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 487 // CHECK1: omp.inner.for.cond: 488 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5:![0-9]+]] 489 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP5]] 490 // CHECK1-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 491 // CHECK1-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 492 // CHECK1: omp.inner.for.body: 493 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 494 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 495 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 496 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP5]] 497 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP5]] 498 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP17]] to i64 499 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], ptr [[TMP2]], i64 0, i64 [[IDXPROM]] 500 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP5]] 501 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 502 // CHECK1: omp.body.continue: 503 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 504 // CHECK1: omp.inner.for.inc: 505 // CHECK1-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 506 // CHECK1-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP18]], 1 507 // CHECK1-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 508 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] 509 // CHECK1: omp.inner.for.end: 510 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 511 // CHECK1: omp.loop.exit: 512 // CHECK1-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 513 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4 514 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP20]]) 515 // CHECK1-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 516 // CHECK1-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 517 // CHECK1-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 518 // CHECK1: .omp.final.then: 519 // CHECK1-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 520 // CHECK1-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP23]], 0 521 // CHECK1-NEXT: [[DIV9:%.*]] = sdiv i32 [[SUB8]], 1 522 // CHECK1-NEXT: [[MUL10:%.*]] = mul nsw i32 [[DIV9]], 1 523 // CHECK1-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]] 524 // CHECK1-NEXT: store i32 [[ADD11]], ptr [[TMP0]], align 4 525 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 526 // CHECK1: .omp.final.done: 527 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 528 // CHECK1: omp.precond.end: 529 // CHECK1-NEXT: ret void 530 // 531 // 532 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l36 533 // CHECK1-SAME: (i64 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 534 // CHECK1-NEXT: entry: 535 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 536 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 537 // CHECK1-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 538 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 539 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 540 // CHECK1-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l36.omp_outlined, ptr [[N_ADDR]], ptr [[TMP0]]) 541 // CHECK1-NEXT: ret void 542 // 543 // 544 // CHECK1-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l36.omp_outlined 545 // CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 546 // CHECK1-NEXT: entry: 547 // CHECK1-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 548 // CHECK1-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 549 // CHECK1-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8 550 // CHECK1-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 551 // CHECK1-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 552 // CHECK1-NEXT: [[TMP:%.*]] = alloca i32, align 4 553 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 554 // CHECK1-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 555 // CHECK1-NEXT: [[I:%.*]] = alloca i32, align 4 556 // CHECK1-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 557 // CHECK1-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 558 // CHECK1-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 559 // CHECK1-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 560 // CHECK1-NEXT: [[I3:%.*]] = alloca i32, align 4 561 // CHECK1-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 562 // CHECK1-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 563 // CHECK1-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8 564 // CHECK1-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 565 // CHECK1-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8 566 // CHECK1-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 567 // CHECK1-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP0]], align 4 568 // CHECK1-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 569 // CHECK1-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 570 // CHECK1-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 571 // CHECK1-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 572 // CHECK1-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 573 // CHECK1-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 574 // CHECK1-NEXT: store i32 0, ptr [[I]], align 4 575 // CHECK1-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 576 // CHECK1-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 577 // CHECK1-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 578 // CHECK1: omp.precond.then: 579 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 580 // CHECK1-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 581 // CHECK1-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_UB]], align 4 582 // CHECK1-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 583 // CHECK1-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 584 // CHECK1-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 585 // CHECK1-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4 586 // CHECK1-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP7]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 587 // CHECK1-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 588 // CHECK1-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 589 // CHECK1-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 590 // CHECK1-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 591 // CHECK1: cond.true: 592 // CHECK1-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 593 // CHECK1-NEXT: br label [[COND_END:%.*]] 594 // CHECK1: cond.false: 595 // CHECK1-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 596 // CHECK1-NEXT: br label [[COND_END]] 597 // CHECK1: cond.end: 598 // CHECK1-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 599 // CHECK1-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 600 // CHECK1-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 601 // CHECK1-NEXT: store i32 [[TMP12]], ptr [[DOTOMP_IV]], align 4 602 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 603 // CHECK1: omp.inner.for.cond: 604 // CHECK1-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 605 // CHECK1-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 606 // CHECK1-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 607 // CHECK1-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 608 // CHECK1: omp.inner.for.body: 609 // CHECK1-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 610 // CHECK1-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP15]], 1 611 // CHECK1-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 612 // CHECK1-NEXT: store i32 [[ADD]], ptr [[I3]], align 4 613 // CHECK1-NEXT: [[TMP16:%.*]] = load i32, ptr [[I3]], align 4 614 // CHECK1-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP16]] to i64 615 // CHECK1-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], ptr [[TMP1]], i64 0, i64 [[IDXPROM]] 616 // CHECK1-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 617 // CHECK1-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 618 // CHECK1: omp.body.continue: 619 // CHECK1-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 620 // CHECK1: omp.inner.for.inc: 621 // CHECK1-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 622 // CHECK1-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP17]], 1 623 // CHECK1-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 624 // CHECK1-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]] 625 // CHECK1: omp.inner.for.end: 626 // CHECK1-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 627 // CHECK1: omp.loop.exit: 628 // CHECK1-NEXT: [[TMP18:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 629 // CHECK1-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP18]], align 4 630 // CHECK1-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP19]]) 631 // CHECK1-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 632 // CHECK1-NEXT: [[TMP21:%.*]] = icmp ne i32 [[TMP20]], 0 633 // CHECK1-NEXT: br i1 [[TMP21]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 634 // CHECK1: .omp.final.then: 635 // CHECK1-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 636 // CHECK1-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP22]], 0 637 // CHECK1-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 638 // CHECK1-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1 639 // CHECK1-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]] 640 // CHECK1-NEXT: store i32 [[ADD10]], ptr [[I3]], align 4 641 // CHECK1-NEXT: br label [[DOTOMP_FINAL_DONE]] 642 // CHECK1: .omp.final.done: 643 // CHECK1-NEXT: br label [[OMP_PRECOND_END]] 644 // CHECK1: omp.precond.end: 645 // CHECK1-NEXT: ret void 646 // 647 // 648 // CHECK3-LABEL: define {{[^@]+}}@_Z21teams_argument_globali 649 // CHECK3-SAME: (i32 noundef [[N:%.*]]) #[[ATTR0:[0-9]+]] { 650 // CHECK3-NEXT: entry: 651 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 652 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 653 // CHECK3-NEXT: [[TE:%.*]] = alloca i32, align 4 654 // CHECK3-NEXT: [[TH:%.*]] = alloca i32, align 4 655 // CHECK3-NEXT: [[TE_CASTED:%.*]] = alloca i32, align 4 656 // CHECK3-NEXT: [[TH_CASTED:%.*]] = alloca i32, align 4 657 // CHECK3-NEXT: [[I_CASTED:%.*]] = alloca i32, align 4 658 // CHECK3-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 659 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [5 x ptr], align 4 660 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [5 x ptr], align 4 661 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [5 x ptr], align 4 662 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 663 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 664 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 665 // CHECK3-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 666 // CHECK3-NEXT: [[N_CASTED4:%.*]] = alloca i32, align 4 667 // CHECK3-NEXT: [[DOTOFFLOAD_BASEPTRS5:%.*]] = alloca [2 x ptr], align 4 668 // CHECK3-NEXT: [[DOTOFFLOAD_PTRS6:%.*]] = alloca [2 x ptr], align 4 669 // CHECK3-NEXT: [[DOTOFFLOAD_MAPPERS7:%.*]] = alloca [2 x ptr], align 4 670 // CHECK3-NEXT: [[_TMP8:%.*]] = alloca i32, align 4 671 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_9:%.*]] = alloca i32, align 4 672 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_10:%.*]] = alloca i32, align 4 673 // CHECK3-NEXT: [[KERNEL_ARGS15:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS]], align 8 674 // CHECK3-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 675 // CHECK3-NEXT: [[TMP0:%.*]] = load i32, ptr [[N_ADDR]], align 4 676 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP0]], 128 677 // CHECK3-NEXT: store i32 [[DIV]], ptr [[TE]], align 4 678 // CHECK3-NEXT: store i32 128, ptr [[TH]], align 4 679 // CHECK3-NEXT: [[TMP1:%.*]] = load i32, ptr [[TE]], align 4 680 // CHECK3-NEXT: store i32 [[TMP1]], ptr [[TE_CASTED]], align 4 681 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TE_CASTED]], align 4 682 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[TH]], align 4 683 // CHECK3-NEXT: store i32 [[TMP3]], ptr [[TH_CASTED]], align 4 684 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[TH_CASTED]], align 4 685 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[I]], align 4 686 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[I_CASTED]], align 4 687 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[I_CASTED]], align 4 688 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[N_ADDR]], align 4 689 // CHECK3-NEXT: store i32 [[TMP7]], ptr [[N_CASTED]], align 4 690 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[N_CASTED]], align 4 691 // CHECK3-NEXT: [[TMP9:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 692 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[TMP9]], align 4 693 // CHECK3-NEXT: [[TMP10:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 694 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[TMP10]], align 4 695 // CHECK3-NEXT: [[TMP11:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 696 // CHECK3-NEXT: store ptr null, ptr [[TMP11]], align 4 697 // CHECK3-NEXT: [[TMP12:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 698 // CHECK3-NEXT: store i32 [[TMP4]], ptr [[TMP12]], align 4 699 // CHECK3-NEXT: [[TMP13:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 700 // CHECK3-NEXT: store i32 [[TMP4]], ptr [[TMP13]], align 4 701 // CHECK3-NEXT: [[TMP14:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 702 // CHECK3-NEXT: store ptr null, ptr [[TMP14]], align 4 703 // CHECK3-NEXT: [[TMP15:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 704 // CHECK3-NEXT: store ptr @a, ptr [[TMP15]], align 4 705 // CHECK3-NEXT: [[TMP16:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 706 // CHECK3-NEXT: store ptr @a, ptr [[TMP16]], align 4 707 // CHECK3-NEXT: [[TMP17:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 708 // CHECK3-NEXT: store ptr null, ptr [[TMP17]], align 4 709 // CHECK3-NEXT: [[TMP18:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 710 // CHECK3-NEXT: store i32 [[TMP6]], ptr [[TMP18]], align 4 711 // CHECK3-NEXT: [[TMP19:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3 712 // CHECK3-NEXT: store i32 [[TMP6]], ptr [[TMP19]], align 4 713 // CHECK3-NEXT: [[TMP20:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 714 // CHECK3-NEXT: store ptr null, ptr [[TMP20]], align 4 715 // CHECK3-NEXT: [[TMP21:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 4 716 // CHECK3-NEXT: store i32 [[TMP8]], ptr [[TMP21]], align 4 717 // CHECK3-NEXT: [[TMP22:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 4 718 // CHECK3-NEXT: store i32 [[TMP8]], ptr [[TMP22]], align 4 719 // CHECK3-NEXT: [[TMP23:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 4 720 // CHECK3-NEXT: store ptr null, ptr [[TMP23]], align 4 721 // CHECK3-NEXT: [[TMP24:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 722 // CHECK3-NEXT: [[TMP25:%.*]] = getelementptr inbounds [5 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 723 // CHECK3-NEXT: [[TMP26:%.*]] = load i32, ptr [[TE]], align 4 724 // CHECK3-NEXT: [[TMP27:%.*]] = load i32, ptr [[TH]], align 4 725 // CHECK3-NEXT: [[TMP28:%.*]] = icmp ult i32 [[TMP27]], 1 726 // CHECK3-NEXT: [[TMP29:%.*]] = select i1 [[TMP28]], i32 [[TMP27]], i32 1 727 // CHECK3-NEXT: [[TMP30:%.*]] = load i32, ptr [[N_ADDR]], align 4 728 // CHECK3-NEXT: store i32 [[TMP30]], ptr [[DOTCAPTURE_EXPR_]], align 4 729 // CHECK3-NEXT: [[TMP31:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 730 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP31]], 0 731 // CHECK3-NEXT: [[DIV2:%.*]] = sdiv i32 [[SUB]], 1 732 // CHECK3-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV2]], 1 733 // CHECK3-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_1]], align 4 734 // CHECK3-NEXT: [[TMP32:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 735 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP32]], 1 736 // CHECK3-NEXT: [[TMP33:%.*]] = zext i32 [[ADD]] to i64 737 // CHECK3-NEXT: [[TMP34:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP26]], 0 738 // CHECK3-NEXT: [[TMP35:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP29]], 0 739 // CHECK3-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 740 // CHECK3-NEXT: store i32 3, ptr [[TMP36]], align 4 741 // CHECK3-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 742 // CHECK3-NEXT: store i32 5, ptr [[TMP37]], align 4 743 // CHECK3-NEXT: [[TMP38:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 744 // CHECK3-NEXT: store ptr [[TMP24]], ptr [[TMP38]], align 4 745 // CHECK3-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 746 // CHECK3-NEXT: store ptr [[TMP25]], ptr [[TMP39]], align 4 747 // CHECK3-NEXT: [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 748 // CHECK3-NEXT: store ptr @.offload_sizes, ptr [[TMP40]], align 4 749 // CHECK3-NEXT: [[TMP41:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 750 // CHECK3-NEXT: store ptr @.offload_maptypes, ptr [[TMP41]], align 4 751 // CHECK3-NEXT: [[TMP42:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 752 // CHECK3-NEXT: store ptr null, ptr [[TMP42]], align 4 753 // CHECK3-NEXT: [[TMP43:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 754 // CHECK3-NEXT: store ptr null, ptr [[TMP43]], align 4 755 // CHECK3-NEXT: [[TMP44:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 756 // CHECK3-NEXT: store i64 [[TMP33]], ptr [[TMP44]], align 8 757 // CHECK3-NEXT: [[TMP45:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 758 // CHECK3-NEXT: store i64 0, ptr [[TMP45]], align 8 759 // CHECK3-NEXT: [[TMP46:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 760 // CHECK3-NEXT: store [3 x i32] [[TMP34]], ptr [[TMP46]], align 4 761 // CHECK3-NEXT: [[TMP47:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 762 // CHECK3-NEXT: store [3 x i32] [[TMP35]], ptr [[TMP47]], align 4 763 // CHECK3-NEXT: [[TMP48:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 764 // CHECK3-NEXT: store i32 0, ptr [[TMP48]], align 4 765 // CHECK3-NEXT: [[TMP49:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 [[TMP26]], i32 [[TMP29]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l30.region_id, ptr [[KERNEL_ARGS]]) 766 // CHECK3-NEXT: [[TMP50:%.*]] = icmp ne i32 [[TMP49]], 0 767 // CHECK3-NEXT: br i1 [[TMP50]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 768 // CHECK3: omp_offload.failed: 769 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l30(i32 [[TMP2]], i32 [[TMP4]], ptr @a, i32 [[TMP6]], i32 [[TMP8]]) #[[ATTR3:[0-9]+]] 770 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT]] 771 // CHECK3: omp_offload.cont: 772 // CHECK3-NEXT: [[TMP51:%.*]] = load i32, ptr [[N_ADDR]], align 4 773 // CHECK3-NEXT: store i32 [[TMP51]], ptr [[N_CASTED4]], align 4 774 // CHECK3-NEXT: [[TMP52:%.*]] = load i32, ptr [[N_CASTED4]], align 4 775 // CHECK3-NEXT: [[TMP53:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 776 // CHECK3-NEXT: store i32 [[TMP52]], ptr [[TMP53]], align 4 777 // CHECK3-NEXT: [[TMP54:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 778 // CHECK3-NEXT: store i32 [[TMP52]], ptr [[TMP54]], align 4 779 // CHECK3-NEXT: [[TMP55:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS7]], i32 0, i32 0 780 // CHECK3-NEXT: store ptr null, ptr [[TMP55]], align 4 781 // CHECK3-NEXT: [[TMP56:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 1 782 // CHECK3-NEXT: store ptr @a, ptr [[TMP56]], align 4 783 // CHECK3-NEXT: [[TMP57:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 1 784 // CHECK3-NEXT: store ptr @a, ptr [[TMP57]], align 4 785 // CHECK3-NEXT: [[TMP58:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_MAPPERS7]], i32 0, i32 1 786 // CHECK3-NEXT: store ptr null, ptr [[TMP58]], align 4 787 // CHECK3-NEXT: [[TMP59:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_BASEPTRS5]], i32 0, i32 0 788 // CHECK3-NEXT: [[TMP60:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOFFLOAD_PTRS6]], i32 0, i32 0 789 // CHECK3-NEXT: [[TMP61:%.*]] = load i32, ptr [[N_ADDR]], align 4 790 // CHECK3-NEXT: store i32 [[TMP61]], ptr [[DOTCAPTURE_EXPR_9]], align 4 791 // CHECK3-NEXT: [[TMP62:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_9]], align 4 792 // CHECK3-NEXT: [[SUB11:%.*]] = sub nsw i32 [[TMP62]], 0 793 // CHECK3-NEXT: [[DIV12:%.*]] = sdiv i32 [[SUB11]], 1 794 // CHECK3-NEXT: [[SUB13:%.*]] = sub nsw i32 [[DIV12]], 1 795 // CHECK3-NEXT: store i32 [[SUB13]], ptr [[DOTCAPTURE_EXPR_10]], align 4 796 // CHECK3-NEXT: [[TMP63:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_10]], align 4 797 // CHECK3-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP63]], 1 798 // CHECK3-NEXT: [[TMP64:%.*]] = zext i32 [[ADD14]] to i64 799 // CHECK3-NEXT: [[TMP65:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 0 800 // CHECK3-NEXT: store i32 3, ptr [[TMP65]], align 4 801 // CHECK3-NEXT: [[TMP66:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 1 802 // CHECK3-NEXT: store i32 2, ptr [[TMP66]], align 4 803 // CHECK3-NEXT: [[TMP67:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 2 804 // CHECK3-NEXT: store ptr [[TMP59]], ptr [[TMP67]], align 4 805 // CHECK3-NEXT: [[TMP68:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 3 806 // CHECK3-NEXT: store ptr [[TMP60]], ptr [[TMP68]], align 4 807 // CHECK3-NEXT: [[TMP69:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 4 808 // CHECK3-NEXT: store ptr @.offload_sizes.1, ptr [[TMP69]], align 4 809 // CHECK3-NEXT: [[TMP70:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 5 810 // CHECK3-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP70]], align 4 811 // CHECK3-NEXT: [[TMP71:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 6 812 // CHECK3-NEXT: store ptr null, ptr [[TMP71]], align 4 813 // CHECK3-NEXT: [[TMP72:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 7 814 // CHECK3-NEXT: store ptr null, ptr [[TMP72]], align 4 815 // CHECK3-NEXT: [[TMP73:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 8 816 // CHECK3-NEXT: store i64 [[TMP64]], ptr [[TMP73]], align 8 817 // CHECK3-NEXT: [[TMP74:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 9 818 // CHECK3-NEXT: store i64 0, ptr [[TMP74]], align 8 819 // CHECK3-NEXT: [[TMP75:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 10 820 // CHECK3-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP75]], align 4 821 // CHECK3-NEXT: [[TMP76:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 11 822 // CHECK3-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP76]], align 4 823 // CHECK3-NEXT: [[TMP77:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS15]], i32 0, i32 12 824 // CHECK3-NEXT: store i32 0, ptr [[TMP77]], align 4 825 // CHECK3-NEXT: [[TMP78:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l36.region_id, ptr [[KERNEL_ARGS15]]) 826 // CHECK3-NEXT: [[TMP79:%.*]] = icmp ne i32 [[TMP78]], 0 827 // CHECK3-NEXT: br i1 [[TMP79]], label [[OMP_OFFLOAD_FAILED16:%.*]], label [[OMP_OFFLOAD_CONT17:%.*]] 828 // CHECK3: omp_offload.failed16: 829 // CHECK3-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l36(i32 [[TMP52]], ptr @a) #[[ATTR3]] 830 // CHECK3-NEXT: br label [[OMP_OFFLOAD_CONT17]] 831 // CHECK3: omp_offload.cont17: 832 // CHECK3-NEXT: [[TMP80:%.*]] = load i32, ptr @a, align 4 833 // CHECK3-NEXT: ret i32 [[TMP80]] 834 // 835 // 836 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l30 837 // CHECK3-SAME: (i32 noundef [[TE:%.*]], i32 noundef [[TH:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]], i32 noundef [[I:%.*]], i32 noundef [[N:%.*]]) #[[ATTR1:[0-9]+]] { 838 // CHECK3-NEXT: entry: 839 // CHECK3-NEXT: [[TE_ADDR:%.*]] = alloca i32, align 4 840 // CHECK3-NEXT: [[TH_ADDR:%.*]] = alloca i32, align 4 841 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 842 // CHECK3-NEXT: [[I_ADDR:%.*]] = alloca i32, align 4 843 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 844 // CHECK3-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2]]) 845 // CHECK3-NEXT: store i32 [[TE]], ptr [[TE_ADDR]], align 4 846 // CHECK3-NEXT: store i32 [[TH]], ptr [[TH_ADDR]], align 4 847 // CHECK3-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 848 // CHECK3-NEXT: store i32 [[I]], ptr [[I_ADDR]], align 4 849 // CHECK3-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 850 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 851 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TE_ADDR]], align 4 852 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[TH_ADDR]], align 4 853 // CHECK3-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB2]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 854 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l30.omp_outlined, ptr [[I_ADDR]], ptr [[N_ADDR]], ptr [[TMP1]]) 855 // CHECK3-NEXT: ret void 856 // 857 // 858 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l30.omp_outlined 859 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[I:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 860 // CHECK3-NEXT: entry: 861 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 862 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 863 // CHECK3-NEXT: [[I_ADDR:%.*]] = alloca ptr, align 4 864 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4 865 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 866 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 867 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 868 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 869 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 870 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 871 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 872 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 873 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 874 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 875 // CHECK3-NEXT: [[I4:%.*]] = alloca i32, align 4 876 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 877 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 878 // CHECK3-NEXT: store ptr [[I]], ptr [[I_ADDR]], align 4 879 // CHECK3-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4 880 // CHECK3-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 881 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[I_ADDR]], align 4 882 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[N_ADDR]], align 4 883 // CHECK3-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4 884 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP1]], align 4 885 // CHECK3-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 886 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 887 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 888 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 889 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 890 // CHECK3-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 891 // CHECK3-NEXT: store i32 0, ptr [[I3]], align 4 892 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 893 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 894 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 895 // CHECK3: omp.precond.then: 896 // CHECK3-NEXT: [[ARRAYDECAY:%.*]] = getelementptr inbounds [100 x i32], ptr [[TMP2]], i32 0, i32 0 897 // CHECK3-NEXT: call void @llvm.assume(i1 true) [ "align"(ptr [[ARRAYDECAY]], i32 16) ] 898 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 899 // CHECK3-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 900 // CHECK3-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 901 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 902 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 903 // CHECK3-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 904 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 905 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 906 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 907 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 908 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 909 // CHECK3-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 910 // CHECK3: cond.true: 911 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 912 // CHECK3-NEXT: br label [[COND_END:%.*]] 913 // CHECK3: cond.false: 914 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 915 // CHECK3-NEXT: br label [[COND_END]] 916 // CHECK3: cond.end: 917 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 918 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 919 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 920 // CHECK3-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 921 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 922 // CHECK3: omp.inner.for.cond: 923 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6:![0-9]+]] 924 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP6]] 925 // CHECK3-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 926 // CHECK3-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 927 // CHECK3: omp.inner.for.body: 928 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]] 929 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 930 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 931 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP6]] 932 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP6]] 933 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], ptr [[TMP2]], i32 0, i32 [[TMP17]] 934 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP6]] 935 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 936 // CHECK3: omp.body.continue: 937 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 938 // CHECK3: omp.inner.for.inc: 939 // CHECK3-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]] 940 // CHECK3-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP18]], 1 941 // CHECK3-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]] 942 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] 943 // CHECK3: omp.inner.for.end: 944 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 945 // CHECK3: omp.loop.exit: 946 // CHECK3-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 947 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4 948 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP20]]) 949 // CHECK3-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 950 // CHECK3-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 951 // CHECK3-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 952 // CHECK3: .omp.final.then: 953 // CHECK3-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 954 // CHECK3-NEXT: [[SUB8:%.*]] = sub nsw i32 [[TMP23]], 0 955 // CHECK3-NEXT: [[DIV9:%.*]] = sdiv i32 [[SUB8]], 1 956 // CHECK3-NEXT: [[MUL10:%.*]] = mul nsw i32 [[DIV9]], 1 957 // CHECK3-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]] 958 // CHECK3-NEXT: store i32 [[ADD11]], ptr [[TMP0]], align 4 959 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 960 // CHECK3: .omp.final.done: 961 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 962 // CHECK3: omp.precond.end: 963 // CHECK3-NEXT: ret void 964 // 965 // 966 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l36 967 // CHECK3-SAME: (i32 noundef [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 968 // CHECK3-NEXT: entry: 969 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 970 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 971 // CHECK3-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 972 // CHECK3-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 973 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 974 // CHECK3-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l36.omp_outlined, ptr [[N_ADDR]], ptr [[TMP0]]) 975 // CHECK3-NEXT: ret void 976 // 977 // 978 // CHECK3-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z21teams_argument_globali_l36.omp_outlined 979 // CHECK3-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], ptr noundef nonnull align 4 dereferenceable(400) [[A:%.*]]) #[[ATTR1]] { 980 // CHECK3-NEXT: entry: 981 // CHECK3-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 982 // CHECK3-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 983 // CHECK3-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4 984 // CHECK3-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 985 // CHECK3-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 986 // CHECK3-NEXT: [[TMP:%.*]] = alloca i32, align 4 987 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 988 // CHECK3-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 989 // CHECK3-NEXT: [[I:%.*]] = alloca i32, align 4 990 // CHECK3-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 991 // CHECK3-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 992 // CHECK3-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 993 // CHECK3-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 994 // CHECK3-NEXT: [[I3:%.*]] = alloca i32, align 4 995 // CHECK3-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 996 // CHECK3-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 997 // CHECK3-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4 998 // CHECK3-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 999 // CHECK3-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4 1000 // CHECK3-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 1001 // CHECK3-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP0]], align 4 1002 // CHECK3-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 1003 // CHECK3-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1004 // CHECK3-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 1005 // CHECK3-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1006 // CHECK3-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1007 // CHECK3-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1008 // CHECK3-NEXT: store i32 0, ptr [[I]], align 4 1009 // CHECK3-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1010 // CHECK3-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 1011 // CHECK3-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1012 // CHECK3: omp.precond.then: 1013 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1014 // CHECK3-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1015 // CHECK3-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_UB]], align 4 1016 // CHECK3-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1017 // CHECK3-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1018 // CHECK3-NEXT: [[TMP6:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1019 // CHECK3-NEXT: [[TMP7:%.*]] = load i32, ptr [[TMP6]], align 4 1020 // CHECK3-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP7]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1021 // CHECK3-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1022 // CHECK3-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1023 // CHECK3-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP8]], [[TMP9]] 1024 // CHECK3-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1025 // CHECK3: cond.true: 1026 // CHECK3-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1027 // CHECK3-NEXT: br label [[COND_END:%.*]] 1028 // CHECK3: cond.false: 1029 // CHECK3-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1030 // CHECK3-NEXT: br label [[COND_END]] 1031 // CHECK3: cond.end: 1032 // CHECK3-NEXT: [[COND:%.*]] = phi i32 [ [[TMP10]], [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ] 1033 // CHECK3-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 1034 // CHECK3-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1035 // CHECK3-NEXT: store i32 [[TMP12]], ptr [[DOTOMP_IV]], align 4 1036 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1037 // CHECK3: omp.inner.for.cond: 1038 // CHECK3-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1039 // CHECK3-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1040 // CHECK3-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 1041 // CHECK3-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1042 // CHECK3: omp.inner.for.body: 1043 // CHECK3-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1044 // CHECK3-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP15]], 1 1045 // CHECK3-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1046 // CHECK3-NEXT: store i32 [[ADD]], ptr [[I3]], align 4 1047 // CHECK3-NEXT: [[TMP16:%.*]] = load i32, ptr [[I3]], align 4 1048 // CHECK3-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], ptr [[TMP1]], i32 0, i32 [[TMP16]] 1049 // CHECK3-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4 1050 // CHECK3-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1051 // CHECK3: omp.body.continue: 1052 // CHECK3-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1053 // CHECK3: omp.inner.for.inc: 1054 // CHECK3-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 1055 // CHECK3-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP17]], 1 1056 // CHECK3-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4 1057 // CHECK3-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP13:![0-9]+]] 1058 // CHECK3: omp.inner.for.end: 1059 // CHECK3-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1060 // CHECK3: omp.loop.exit: 1061 // CHECK3-NEXT: [[TMP18:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1062 // CHECK3-NEXT: [[TMP19:%.*]] = load i32, ptr [[TMP18]], align 4 1063 // CHECK3-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP19]]) 1064 // CHECK3-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 1065 // CHECK3-NEXT: [[TMP21:%.*]] = icmp ne i32 [[TMP20]], 0 1066 // CHECK3-NEXT: br i1 [[TMP21]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 1067 // CHECK3: .omp.final.then: 1068 // CHECK3-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1069 // CHECK3-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP22]], 0 1070 // CHECK3-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 1071 // CHECK3-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1 1072 // CHECK3-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]] 1073 // CHECK3-NEXT: store i32 [[ADD10]], ptr [[I3]], align 4 1074 // CHECK3-NEXT: br label [[DOTOMP_FINAL_DONE]] 1075 // CHECK3: .omp.final.done: 1076 // CHECK3-NEXT: br label [[OMP_PRECOND_END]] 1077 // CHECK3: omp.precond.end: 1078 // CHECK3-NEXT: ret void 1079 // 1080 // 1081 // CHECK5-LABEL: define {{[^@]+}}@_Z21teams_argument_globali 1082 // CHECK5-SAME: (i32 noundef signext [[N:%.*]]) #[[ATTR0:[0-9]+]] { 1083 // CHECK5-NEXT: entry: 1084 // CHECK5-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 1085 // CHECK5-NEXT: [[I:%.*]] = alloca i32, align 4 1086 // CHECK5-NEXT: [[TE:%.*]] = alloca i32, align 4 1087 // CHECK5-NEXT: [[TH:%.*]] = alloca i32, align 4 1088 // CHECK5-NEXT: [[TMP:%.*]] = alloca i32, align 4 1089 // CHECK5-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1090 // CHECK5-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1091 // CHECK5-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1092 // CHECK5-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1093 // CHECK5-NEXT: [[I4:%.*]] = alloca i32, align 4 1094 // CHECK5-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1095 // CHECK5-NEXT: [[DOTLINEAR_START:%.*]] = alloca i32, align 4 1096 // CHECK5-NEXT: [[I5:%.*]] = alloca i32, align 4 1097 // CHECK5-NEXT: [[I6:%.*]] = alloca i32, align 4 1098 // CHECK5-NEXT: [[_TMP13:%.*]] = alloca i32, align 4 1099 // CHECK5-NEXT: [[DOTCAPTURE_EXPR_14:%.*]] = alloca i32, align 4 1100 // CHECK5-NEXT: [[DOTCAPTURE_EXPR_15:%.*]] = alloca i32, align 4 1101 // CHECK5-NEXT: [[DOTOMP_LB19:%.*]] = alloca i32, align 4 1102 // CHECK5-NEXT: [[DOTOMP_UB20:%.*]] = alloca i32, align 4 1103 // CHECK5-NEXT: [[I21:%.*]] = alloca i32, align 4 1104 // CHECK5-NEXT: [[DOTOMP_IV24:%.*]] = alloca i32, align 4 1105 // CHECK5-NEXT: [[I25:%.*]] = alloca i32, align 4 1106 // CHECK5-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 1107 // CHECK5-NEXT: [[TMP0:%.*]] = load i32, ptr [[N_ADDR]], align 4 1108 // CHECK5-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP0]], 128 1109 // CHECK5-NEXT: store i32 [[DIV]], ptr [[TE]], align 4 1110 // CHECK5-NEXT: store i32 128, ptr [[TH]], align 4 1111 // CHECK5-NEXT: [[TMP1:%.*]] = load i32, ptr [[N_ADDR]], align 4 1112 // CHECK5-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4 1113 // CHECK5-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1114 // CHECK5-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP2]], 0 1115 // CHECK5-NEXT: [[DIV2:%.*]] = sdiv i32 [[SUB]], 1 1116 // CHECK5-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV2]], 1 1117 // CHECK5-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1118 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1119 // CHECK5-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1120 // CHECK5-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_UB]], align 4 1121 // CHECK5-NEXT: store i32 0, ptr [[I4]], align 4 1122 // CHECK5-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1123 // CHECK5-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 1124 // CHECK5-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 1125 // CHECK5: simd.if.then: 1126 // CHECK5-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1127 // CHECK5-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 1128 // CHECK5-NEXT: call void @llvm.assume(i1 true) [ "align"(ptr @a, i64 16) ] 1129 // CHECK5-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4 1130 // CHECK5-NEXT: store i32 [[TMP6]], ptr [[DOTLINEAR_START]], align 4 1131 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1132 // CHECK5: omp.inner.for.cond: 1133 // CHECK5-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2:![0-9]+]] 1134 // CHECK5-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP2]] 1135 // CHECK5-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] 1136 // CHECK5-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1137 // CHECK5: omp.inner.for.body: 1138 // CHECK5-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 1139 // CHECK5-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 1140 // CHECK5-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1141 // CHECK5-NEXT: store i32 [[ADD]], ptr [[I5]], align 4, !llvm.access.group [[ACC_GRP2]] 1142 // CHECK5-NEXT: [[TMP10:%.*]] = load i32, ptr [[I5]], align 4, !llvm.access.group [[ACC_GRP2]] 1143 // CHECK5-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP10]] to i64 1144 // CHECK5-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], ptr @a, i64 0, i64 [[IDXPROM]] 1145 // CHECK5-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP2]] 1146 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1147 // CHECK5: omp.body.continue: 1148 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1149 // CHECK5: omp.inner.for.inc: 1150 // CHECK5-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 1151 // CHECK5-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP11]], 1 1152 // CHECK5-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 1153 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] 1154 // CHECK5: omp.inner.for.end: 1155 // CHECK5-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1156 // CHECK5-NEXT: [[SUB9:%.*]] = sub nsw i32 [[TMP12]], 0 1157 // CHECK5-NEXT: [[DIV10:%.*]] = sdiv i32 [[SUB9]], 1 1158 // CHECK5-NEXT: [[MUL11:%.*]] = mul nsw i32 [[DIV10]], 1 1159 // CHECK5-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]] 1160 // CHECK5-NEXT: store i32 [[ADD12]], ptr [[I]], align 4 1161 // CHECK5-NEXT: br label [[SIMD_IF_END]] 1162 // CHECK5: simd.if.end: 1163 // CHECK5-NEXT: [[TMP13:%.*]] = load i32, ptr [[N_ADDR]], align 4 1164 // CHECK5-NEXT: store i32 [[TMP13]], ptr [[DOTCAPTURE_EXPR_14]], align 4 1165 // CHECK5-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_14]], align 4 1166 // CHECK5-NEXT: [[SUB16:%.*]] = sub nsw i32 [[TMP14]], 0 1167 // CHECK5-NEXT: [[DIV17:%.*]] = sdiv i32 [[SUB16]], 1 1168 // CHECK5-NEXT: [[SUB18:%.*]] = sub nsw i32 [[DIV17]], 1 1169 // CHECK5-NEXT: store i32 [[SUB18]], ptr [[DOTCAPTURE_EXPR_15]], align 4 1170 // CHECK5-NEXT: store i32 0, ptr [[DOTOMP_LB19]], align 4 1171 // CHECK5-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_15]], align 4 1172 // CHECK5-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_UB20]], align 4 1173 // CHECK5-NEXT: store i32 0, ptr [[I21]], align 4 1174 // CHECK5-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_14]], align 4 1175 // CHECK5-NEXT: [[CMP22:%.*]] = icmp slt i32 0, [[TMP16]] 1176 // CHECK5-NEXT: br i1 [[CMP22]], label [[SIMD_IF_THEN23:%.*]], label [[SIMD_IF_END41:%.*]] 1177 // CHECK5: simd.if.then23: 1178 // CHECK5-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_LB19]], align 4 1179 // CHECK5-NEXT: store i32 [[TMP17]], ptr [[DOTOMP_IV24]], align 4 1180 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND26:%.*]] 1181 // CHECK5: omp.inner.for.cond26: 1182 // CHECK5-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV24]], align 4 1183 // CHECK5-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_UB20]], align 4 1184 // CHECK5-NEXT: [[CMP27:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]] 1185 // CHECK5-NEXT: br i1 [[CMP27]], label [[OMP_INNER_FOR_BODY28:%.*]], label [[OMP_INNER_FOR_END36:%.*]] 1186 // CHECK5: omp.inner.for.body28: 1187 // CHECK5-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV24]], align 4 1188 // CHECK5-NEXT: [[MUL29:%.*]] = mul nsw i32 [[TMP20]], 1 1189 // CHECK5-NEXT: [[ADD30:%.*]] = add nsw i32 0, [[MUL29]] 1190 // CHECK5-NEXT: store i32 [[ADD30]], ptr [[I25]], align 4 1191 // CHECK5-NEXT: [[TMP21:%.*]] = load i32, ptr [[I25]], align 4 1192 // CHECK5-NEXT: [[IDXPROM31:%.*]] = sext i32 [[TMP21]] to i64 1193 // CHECK5-NEXT: [[ARRAYIDX32:%.*]] = getelementptr inbounds [100 x i32], ptr @a, i64 0, i64 [[IDXPROM31]] 1194 // CHECK5-NEXT: store i32 0, ptr [[ARRAYIDX32]], align 4 1195 // CHECK5-NEXT: br label [[OMP_BODY_CONTINUE33:%.*]] 1196 // CHECK5: omp.body.continue33: 1197 // CHECK5-NEXT: br label [[OMP_INNER_FOR_INC34:%.*]] 1198 // CHECK5: omp.inner.for.inc34: 1199 // CHECK5-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IV24]], align 4 1200 // CHECK5-NEXT: [[ADD35:%.*]] = add nsw i32 [[TMP22]], 1 1201 // CHECK5-NEXT: store i32 [[ADD35]], ptr [[DOTOMP_IV24]], align 4 1202 // CHECK5-NEXT: br label [[OMP_INNER_FOR_COND26]], !llvm.loop [[LOOP7:![0-9]+]] 1203 // CHECK5: omp.inner.for.end36: 1204 // CHECK5-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_14]], align 4 1205 // CHECK5-NEXT: [[SUB37:%.*]] = sub nsw i32 [[TMP23]], 0 1206 // CHECK5-NEXT: [[DIV38:%.*]] = sdiv i32 [[SUB37]], 1 1207 // CHECK5-NEXT: [[MUL39:%.*]] = mul nsw i32 [[DIV38]], 1 1208 // CHECK5-NEXT: [[ADD40:%.*]] = add nsw i32 0, [[MUL39]] 1209 // CHECK5-NEXT: store i32 [[ADD40]], ptr [[I25]], align 4 1210 // CHECK5-NEXT: br label [[SIMD_IF_END41]] 1211 // CHECK5: simd.if.end41: 1212 // CHECK5-NEXT: [[TMP24:%.*]] = load i32, ptr @a, align 4 1213 // CHECK5-NEXT: ret i32 [[TMP24]] 1214 // 1215 // 1216 // CHECK7-LABEL: define {{[^@]+}}@_Z21teams_argument_globali 1217 // CHECK7-SAME: (i32 noundef [[N:%.*]]) #[[ATTR0:[0-9]+]] { 1218 // CHECK7-NEXT: entry: 1219 // CHECK7-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 1220 // CHECK7-NEXT: [[I:%.*]] = alloca i32, align 4 1221 // CHECK7-NEXT: [[TE:%.*]] = alloca i32, align 4 1222 // CHECK7-NEXT: [[TH:%.*]] = alloca i32, align 4 1223 // CHECK7-NEXT: [[TMP:%.*]] = alloca i32, align 4 1224 // CHECK7-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1225 // CHECK7-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1226 // CHECK7-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1227 // CHECK7-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1228 // CHECK7-NEXT: [[I4:%.*]] = alloca i32, align 4 1229 // CHECK7-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1230 // CHECK7-NEXT: [[DOTLINEAR_START:%.*]] = alloca i32, align 4 1231 // CHECK7-NEXT: [[I5:%.*]] = alloca i32, align 4 1232 // CHECK7-NEXT: [[I6:%.*]] = alloca i32, align 4 1233 // CHECK7-NEXT: [[_TMP13:%.*]] = alloca i32, align 4 1234 // CHECK7-NEXT: [[DOTCAPTURE_EXPR_14:%.*]] = alloca i32, align 4 1235 // CHECK7-NEXT: [[DOTCAPTURE_EXPR_15:%.*]] = alloca i32, align 4 1236 // CHECK7-NEXT: [[DOTOMP_LB19:%.*]] = alloca i32, align 4 1237 // CHECK7-NEXT: [[DOTOMP_UB20:%.*]] = alloca i32, align 4 1238 // CHECK7-NEXT: [[I21:%.*]] = alloca i32, align 4 1239 // CHECK7-NEXT: [[DOTOMP_IV24:%.*]] = alloca i32, align 4 1240 // CHECK7-NEXT: [[I25:%.*]] = alloca i32, align 4 1241 // CHECK7-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 1242 // CHECK7-NEXT: [[TMP0:%.*]] = load i32, ptr [[N_ADDR]], align 4 1243 // CHECK7-NEXT: [[DIV:%.*]] = sdiv i32 [[TMP0]], 128 1244 // CHECK7-NEXT: store i32 [[DIV]], ptr [[TE]], align 4 1245 // CHECK7-NEXT: store i32 128, ptr [[TH]], align 4 1246 // CHECK7-NEXT: [[TMP1:%.*]] = load i32, ptr [[N_ADDR]], align 4 1247 // CHECK7-NEXT: store i32 [[TMP1]], ptr [[DOTCAPTURE_EXPR_]], align 4 1248 // CHECK7-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1249 // CHECK7-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP2]], 0 1250 // CHECK7-NEXT: [[DIV2:%.*]] = sdiv i32 [[SUB]], 1 1251 // CHECK7-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV2]], 1 1252 // CHECK7-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1253 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1254 // CHECK7-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1255 // CHECK7-NEXT: store i32 [[TMP3]], ptr [[DOTOMP_UB]], align 4 1256 // CHECK7-NEXT: store i32 0, ptr [[I4]], align 4 1257 // CHECK7-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1258 // CHECK7-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP4]] 1259 // CHECK7-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 1260 // CHECK7: simd.if.then: 1261 // CHECK7-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1262 // CHECK7-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 1263 // CHECK7-NEXT: call void @llvm.assume(i1 true) [ "align"(ptr @a, i32 16) ] 1264 // CHECK7-NEXT: [[TMP6:%.*]] = load i32, ptr [[I]], align 4 1265 // CHECK7-NEXT: store i32 [[TMP6]], ptr [[DOTLINEAR_START]], align 4 1266 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1267 // CHECK7: omp.inner.for.cond: 1268 // CHECK7-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3:![0-9]+]] 1269 // CHECK7-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP3]] 1270 // CHECK7-NEXT: [[CMP7:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] 1271 // CHECK7-NEXT: br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1272 // CHECK7: omp.inner.for.body: 1273 // CHECK7-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 1274 // CHECK7-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 1275 // CHECK7-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1276 // CHECK7-NEXT: store i32 [[ADD]], ptr [[I5]], align 4, !llvm.access.group [[ACC_GRP3]] 1277 // CHECK7-NEXT: [[TMP10:%.*]] = load i32, ptr [[I5]], align 4, !llvm.access.group [[ACC_GRP3]] 1278 // CHECK7-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [100 x i32], ptr @a, i32 0, i32 [[TMP10]] 1279 // CHECK7-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP3]] 1280 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1281 // CHECK7: omp.body.continue: 1282 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1283 // CHECK7: omp.inner.for.inc: 1284 // CHECK7-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 1285 // CHECK7-NEXT: [[ADD8:%.*]] = add nsw i32 [[TMP11]], 1 1286 // CHECK7-NEXT: store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 1287 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] 1288 // CHECK7: omp.inner.for.end: 1289 // CHECK7-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1290 // CHECK7-NEXT: [[SUB9:%.*]] = sub nsw i32 [[TMP12]], 0 1291 // CHECK7-NEXT: [[DIV10:%.*]] = sdiv i32 [[SUB9]], 1 1292 // CHECK7-NEXT: [[MUL11:%.*]] = mul nsw i32 [[DIV10]], 1 1293 // CHECK7-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]] 1294 // CHECK7-NEXT: store i32 [[ADD12]], ptr [[I]], align 4 1295 // CHECK7-NEXT: br label [[SIMD_IF_END]] 1296 // CHECK7: simd.if.end: 1297 // CHECK7-NEXT: [[TMP13:%.*]] = load i32, ptr [[N_ADDR]], align 4 1298 // CHECK7-NEXT: store i32 [[TMP13]], ptr [[DOTCAPTURE_EXPR_14]], align 4 1299 // CHECK7-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_14]], align 4 1300 // CHECK7-NEXT: [[SUB16:%.*]] = sub nsw i32 [[TMP14]], 0 1301 // CHECK7-NEXT: [[DIV17:%.*]] = sdiv i32 [[SUB16]], 1 1302 // CHECK7-NEXT: [[SUB18:%.*]] = sub nsw i32 [[DIV17]], 1 1303 // CHECK7-NEXT: store i32 [[SUB18]], ptr [[DOTCAPTURE_EXPR_15]], align 4 1304 // CHECK7-NEXT: store i32 0, ptr [[DOTOMP_LB19]], align 4 1305 // CHECK7-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_15]], align 4 1306 // CHECK7-NEXT: store i32 [[TMP15]], ptr [[DOTOMP_UB20]], align 4 1307 // CHECK7-NEXT: store i32 0, ptr [[I21]], align 4 1308 // CHECK7-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_14]], align 4 1309 // CHECK7-NEXT: [[CMP22:%.*]] = icmp slt i32 0, [[TMP16]] 1310 // CHECK7-NEXT: br i1 [[CMP22]], label [[SIMD_IF_THEN23:%.*]], label [[SIMD_IF_END40:%.*]] 1311 // CHECK7: simd.if.then23: 1312 // CHECK7-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_LB19]], align 4 1313 // CHECK7-NEXT: store i32 [[TMP17]], ptr [[DOTOMP_IV24]], align 4 1314 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND26:%.*]] 1315 // CHECK7: omp.inner.for.cond26: 1316 // CHECK7-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV24]], align 4 1317 // CHECK7-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_UB20]], align 4 1318 // CHECK7-NEXT: [[CMP27:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]] 1319 // CHECK7-NEXT: br i1 [[CMP27]], label [[OMP_INNER_FOR_BODY28:%.*]], label [[OMP_INNER_FOR_END35:%.*]] 1320 // CHECK7: omp.inner.for.body28: 1321 // CHECK7-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV24]], align 4 1322 // CHECK7-NEXT: [[MUL29:%.*]] = mul nsw i32 [[TMP20]], 1 1323 // CHECK7-NEXT: [[ADD30:%.*]] = add nsw i32 0, [[MUL29]] 1324 // CHECK7-NEXT: store i32 [[ADD30]], ptr [[I25]], align 4 1325 // CHECK7-NEXT: [[TMP21:%.*]] = load i32, ptr [[I25]], align 4 1326 // CHECK7-NEXT: [[ARRAYIDX31:%.*]] = getelementptr inbounds [100 x i32], ptr @a, i32 0, i32 [[TMP21]] 1327 // CHECK7-NEXT: store i32 0, ptr [[ARRAYIDX31]], align 4 1328 // CHECK7-NEXT: br label [[OMP_BODY_CONTINUE32:%.*]] 1329 // CHECK7: omp.body.continue32: 1330 // CHECK7-NEXT: br label [[OMP_INNER_FOR_INC33:%.*]] 1331 // CHECK7: omp.inner.for.inc33: 1332 // CHECK7-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IV24]], align 4 1333 // CHECK7-NEXT: [[ADD34:%.*]] = add nsw i32 [[TMP22]], 1 1334 // CHECK7-NEXT: store i32 [[ADD34]], ptr [[DOTOMP_IV24]], align 4 1335 // CHECK7-NEXT: br label [[OMP_INNER_FOR_COND26]], !llvm.loop [[LOOP8:![0-9]+]] 1336 // CHECK7: omp.inner.for.end35: 1337 // CHECK7-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_14]], align 4 1338 // CHECK7-NEXT: [[SUB36:%.*]] = sub nsw i32 [[TMP23]], 0 1339 // CHECK7-NEXT: [[DIV37:%.*]] = sdiv i32 [[SUB36]], 1 1340 // CHECK7-NEXT: [[MUL38:%.*]] = mul nsw i32 [[DIV37]], 1 1341 // CHECK7-NEXT: [[ADD39:%.*]] = add nsw i32 0, [[MUL38]] 1342 // CHECK7-NEXT: store i32 [[ADD39]], ptr [[I25]], align 4 1343 // CHECK7-NEXT: br label [[SIMD_IF_END40]] 1344 // CHECK7: simd.if.end40: 1345 // CHECK7-NEXT: [[TMP24:%.*]] = load i32, ptr @a, align 4 1346 // CHECK7-NEXT: ret i32 [[TMP24]] 1347 // 1348 // 1349 // CHECK9-LABEL: define {{[^@]+}}@_Z15teams_local_argv 1350 // CHECK9-SAME: () #[[ATTR0:[0-9]+]] { 1351 // CHECK9-NEXT: entry: 1352 // CHECK9-NEXT: [[N:%.*]] = alloca i32, align 4 1353 // CHECK9-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 8 1354 // CHECK9-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 1355 // CHECK9-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 1356 // CHECK9-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8 1357 // CHECK9-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8 1358 // CHECK9-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8 1359 // CHECK9-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8 1360 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1361 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1362 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1363 // CHECK9-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 1364 // CHECK9-NEXT: store i32 100, ptr [[N]], align 4 1365 // CHECK9-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 1366 // CHECK9-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 1367 // CHECK9-NEXT: [[TMP2:%.*]] = call ptr @llvm.stacksave.p0() 1368 // CHECK9-NEXT: store ptr [[TMP2]], ptr [[SAVED_STACK]], align 8 1369 // CHECK9-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 1370 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8 1371 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[N]], align 4 1372 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 1373 // CHECK9-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8 1374 // CHECK9-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 1375 // CHECK9-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 24, i1 false) 1376 // CHECK9-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1377 // CHECK9-NEXT: store i64 [[TMP4]], ptr [[TMP6]], align 8 1378 // CHECK9-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1379 // CHECK9-NEXT: store i64 [[TMP4]], ptr [[TMP7]], align 8 1380 // CHECK9-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 1381 // CHECK9-NEXT: store ptr null, ptr [[TMP8]], align 8 1382 // CHECK9-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 1383 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP9]], align 8 1384 // CHECK9-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 1385 // CHECK9-NEXT: store i64 [[TMP1]], ptr [[TMP10]], align 8 1386 // CHECK9-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 1387 // CHECK9-NEXT: store ptr null, ptr [[TMP11]], align 8 1388 // CHECK9-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 1389 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP12]], align 8 1390 // CHECK9-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 1391 // CHECK9-NEXT: store ptr [[VLA]], ptr [[TMP13]], align 8 1392 // CHECK9-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2 1393 // CHECK9-NEXT: store i64 [[TMP5]], ptr [[TMP14]], align 8 1394 // CHECK9-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 1395 // CHECK9-NEXT: store ptr null, ptr [[TMP15]], align 8 1396 // CHECK9-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1397 // CHECK9-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1398 // CHECK9-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 1399 // CHECK9-NEXT: [[TMP19:%.*]] = load i32, ptr [[N]], align 4 1400 // CHECK9-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4 1401 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1402 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0 1403 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1404 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1405 // CHECK9-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1406 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1407 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], 1 1408 // CHECK9-NEXT: [[TMP22:%.*]] = zext i32 [[ADD]] to i64 1409 // CHECK9-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 1410 // CHECK9-NEXT: store i32 3, ptr [[TMP23]], align 4 1411 // CHECK9-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 1412 // CHECK9-NEXT: store i32 3, ptr [[TMP24]], align 4 1413 // CHECK9-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 1414 // CHECK9-NEXT: store ptr [[TMP16]], ptr [[TMP25]], align 8 1415 // CHECK9-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 1416 // CHECK9-NEXT: store ptr [[TMP17]], ptr [[TMP26]], align 8 1417 // CHECK9-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 1418 // CHECK9-NEXT: store ptr [[TMP18]], ptr [[TMP27]], align 8 1419 // CHECK9-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 1420 // CHECK9-NEXT: store ptr @.offload_maptypes, ptr [[TMP28]], align 8 1421 // CHECK9-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 1422 // CHECK9-NEXT: store ptr null, ptr [[TMP29]], align 8 1423 // CHECK9-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 1424 // CHECK9-NEXT: store ptr null, ptr [[TMP30]], align 8 1425 // CHECK9-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 1426 // CHECK9-NEXT: store i64 [[TMP22]], ptr [[TMP31]], align 8 1427 // CHECK9-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 1428 // CHECK9-NEXT: store i64 0, ptr [[TMP32]], align 8 1429 // CHECK9-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 1430 // CHECK9-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4 1431 // CHECK9-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 1432 // CHECK9-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP34]], align 4 1433 // CHECK9-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 1434 // CHECK9-NEXT: store i32 0, ptr [[TMP35]], align 4 1435 // CHECK9-NEXT: [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75.region_id, ptr [[KERNEL_ARGS]]) 1436 // CHECK9-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0 1437 // CHECK9-NEXT: br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1438 // CHECK9: omp_offload.failed: 1439 // CHECK9-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75(i64 [[TMP4]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3:[0-9]+]] 1440 // CHECK9-NEXT: br label [[OMP_OFFLOAD_CONT]] 1441 // CHECK9: omp_offload.cont: 1442 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i64 0 1443 // CHECK9-NEXT: [[TMP38:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 1444 // CHECK9-NEXT: [[TMP39:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 1445 // CHECK9-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP39]]) 1446 // CHECK9-NEXT: ret i32 [[TMP38]] 1447 // 1448 // 1449 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75 1450 // CHECK9-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 1451 // CHECK9-NEXT: entry: 1452 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 1453 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1454 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1455 // CHECK9-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 1456 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 1457 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1458 // CHECK9-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 1459 // CHECK9-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1460 // CHECK9-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]]) 1461 // CHECK9-NEXT: ret void 1462 // 1463 // 1464 // CHECK9-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75.omp_outlined 1465 // CHECK9-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1466 // CHECK9-NEXT: entry: 1467 // CHECK9-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 1468 // CHECK9-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 1469 // CHECK9-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8 1470 // CHECK9-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 1471 // CHECK9-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 1472 // CHECK9-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1473 // CHECK9-NEXT: [[TMP:%.*]] = alloca i32, align 4 1474 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1475 // CHECK9-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1476 // CHECK9-NEXT: [[I:%.*]] = alloca i32, align 4 1477 // CHECK9-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1478 // CHECK9-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1479 // CHECK9-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1480 // CHECK9-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1481 // CHECK9-NEXT: [[I3:%.*]] = alloca i32, align 4 1482 // CHECK9-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 1483 // CHECK9-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 1484 // CHECK9-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8 1485 // CHECK9-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 1486 // CHECK9-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 1487 // CHECK9-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8 1488 // CHECK9-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 1489 // CHECK9-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 1490 // CHECK9-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 1491 // CHECK9-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 1492 // CHECK9-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1493 // CHECK9-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1494 // CHECK9-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1495 // CHECK9-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1496 // CHECK9-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1497 // CHECK9-NEXT: store i32 0, ptr [[I]], align 4 1498 // CHECK9-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1499 // CHECK9-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1500 // CHECK9-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1501 // CHECK9: omp.precond.then: 1502 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1503 // CHECK9-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1504 // CHECK9-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 1505 // CHECK9-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1506 // CHECK9-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1507 // CHECK9-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1508 // CHECK9-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 1509 // CHECK9-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1510 // CHECK9-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1511 // CHECK9-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1512 // CHECK9-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 1513 // CHECK9-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1514 // CHECK9: cond.true: 1515 // CHECK9-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1516 // CHECK9-NEXT: br label [[COND_END:%.*]] 1517 // CHECK9: cond.false: 1518 // CHECK9-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1519 // CHECK9-NEXT: br label [[COND_END]] 1520 // CHECK9: cond.end: 1521 // CHECK9-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 1522 // CHECK9-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 1523 // CHECK9-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1524 // CHECK9-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 1525 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1526 // CHECK9: omp.inner.for.cond: 1527 // CHECK9-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4:![0-9]+]] 1528 // CHECK9-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP4]] 1529 // CHECK9-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 1530 // CHECK9-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1531 // CHECK9: omp.inner.for.body: 1532 // CHECK9-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]] 1533 // CHECK9-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 1534 // CHECK9-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1535 // CHECK9-NEXT: store i32 [[ADD]], ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP4]] 1536 // CHECK9-NEXT: [[TMP17:%.*]] = load i32, ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP4]] 1537 // CHECK9-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP17]] to i64 1538 // CHECK9-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]] 1539 // CHECK9-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP4]] 1540 // CHECK9-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1541 // CHECK9: omp.body.continue: 1542 // CHECK9-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1543 // CHECK9: omp.inner.for.inc: 1544 // CHECK9-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]] 1545 // CHECK9-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], 1 1546 // CHECK9-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]] 1547 // CHECK9-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]] 1548 // CHECK9: omp.inner.for.end: 1549 // CHECK9-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1550 // CHECK9: omp.loop.exit: 1551 // CHECK9-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 1552 // CHECK9-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4 1553 // CHECK9-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP20]]) 1554 // CHECK9-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 1555 // CHECK9-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 1556 // CHECK9-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 1557 // CHECK9: .omp.final.then: 1558 // CHECK9-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1559 // CHECK9-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP23]], 0 1560 // CHECK9-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 1561 // CHECK9-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1 1562 // CHECK9-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]] 1563 // CHECK9-NEXT: store i32 [[ADD10]], ptr [[I3]], align 4 1564 // CHECK9-NEXT: br label [[DOTOMP_FINAL_DONE]] 1565 // CHECK9: .omp.final.done: 1566 // CHECK9-NEXT: br label [[OMP_PRECOND_END]] 1567 // CHECK9: omp.precond.end: 1568 // CHECK9-NEXT: ret void 1569 // 1570 // 1571 // CHECK11-LABEL: define {{[^@]+}}@_Z15teams_local_argv 1572 // CHECK11-SAME: () #[[ATTR0:[0-9]+]] { 1573 // CHECK11-NEXT: entry: 1574 // CHECK11-NEXT: [[N:%.*]] = alloca i32, align 4 1575 // CHECK11-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 4 1576 // CHECK11-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 1577 // CHECK11-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 1578 // CHECK11-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4 1579 // CHECK11-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4 1580 // CHECK11-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4 1581 // CHECK11-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4 1582 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 1583 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1584 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1585 // CHECK11-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 1586 // CHECK11-NEXT: store i32 100, ptr [[N]], align 4 1587 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 1588 // CHECK11-NEXT: [[TMP1:%.*]] = call ptr @llvm.stacksave.p0() 1589 // CHECK11-NEXT: store ptr [[TMP1]], ptr [[SAVED_STACK]], align 4 1590 // CHECK11-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 1591 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[__VLA_EXPR0]], align 4 1592 // CHECK11-NEXT: [[TMP2:%.*]] = load i32, ptr [[N]], align 4 1593 // CHECK11-NEXT: store i32 [[TMP2]], ptr [[N_CASTED]], align 4 1594 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_CASTED]], align 4 1595 // CHECK11-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 1596 // CHECK11-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 1597 // CHECK11-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 24, i1 false) 1598 // CHECK11-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1599 // CHECK11-NEXT: store i32 [[TMP3]], ptr [[TMP6]], align 4 1600 // CHECK11-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1601 // CHECK11-NEXT: store i32 [[TMP3]], ptr [[TMP7]], align 4 1602 // CHECK11-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 1603 // CHECK11-NEXT: store ptr null, ptr [[TMP8]], align 4 1604 // CHECK11-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 1605 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP9]], align 4 1606 // CHECK11-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 1607 // CHECK11-NEXT: store i32 [[TMP0]], ptr [[TMP10]], align 4 1608 // CHECK11-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 1609 // CHECK11-NEXT: store ptr null, ptr [[TMP11]], align 4 1610 // CHECK11-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 1611 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP12]], align 4 1612 // CHECK11-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 1613 // CHECK11-NEXT: store ptr [[VLA]], ptr [[TMP13]], align 4 1614 // CHECK11-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2 1615 // CHECK11-NEXT: store i64 [[TMP5]], ptr [[TMP14]], align 4 1616 // CHECK11-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 1617 // CHECK11-NEXT: store ptr null, ptr [[TMP15]], align 4 1618 // CHECK11-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1619 // CHECK11-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1620 // CHECK11-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 1621 // CHECK11-NEXT: [[TMP19:%.*]] = load i32, ptr [[N]], align 4 1622 // CHECK11-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4 1623 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1624 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0 1625 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1626 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1627 // CHECK11-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1628 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1629 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], 1 1630 // CHECK11-NEXT: [[TMP22:%.*]] = zext i32 [[ADD]] to i64 1631 // CHECK11-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 1632 // CHECK11-NEXT: store i32 3, ptr [[TMP23]], align 4 1633 // CHECK11-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 1634 // CHECK11-NEXT: store i32 3, ptr [[TMP24]], align 4 1635 // CHECK11-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 1636 // CHECK11-NEXT: store ptr [[TMP16]], ptr [[TMP25]], align 4 1637 // CHECK11-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 1638 // CHECK11-NEXT: store ptr [[TMP17]], ptr [[TMP26]], align 4 1639 // CHECK11-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 1640 // CHECK11-NEXT: store ptr [[TMP18]], ptr [[TMP27]], align 4 1641 // CHECK11-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 1642 // CHECK11-NEXT: store ptr @.offload_maptypes, ptr [[TMP28]], align 4 1643 // CHECK11-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 1644 // CHECK11-NEXT: store ptr null, ptr [[TMP29]], align 4 1645 // CHECK11-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 1646 // CHECK11-NEXT: store ptr null, ptr [[TMP30]], align 4 1647 // CHECK11-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 1648 // CHECK11-NEXT: store i64 [[TMP22]], ptr [[TMP31]], align 8 1649 // CHECK11-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 1650 // CHECK11-NEXT: store i64 0, ptr [[TMP32]], align 8 1651 // CHECK11-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 1652 // CHECK11-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4 1653 // CHECK11-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 1654 // CHECK11-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP34]], align 4 1655 // CHECK11-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 1656 // CHECK11-NEXT: store i32 0, ptr [[TMP35]], align 4 1657 // CHECK11-NEXT: [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75.region_id, ptr [[KERNEL_ARGS]]) 1658 // CHECK11-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0 1659 // CHECK11-NEXT: br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 1660 // CHECK11: omp_offload.failed: 1661 // CHECK11-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75(i32 [[TMP3]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3:[0-9]+]] 1662 // CHECK11-NEXT: br label [[OMP_OFFLOAD_CONT]] 1663 // CHECK11: omp_offload.cont: 1664 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i32 0 1665 // CHECK11-NEXT: [[TMP38:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 1666 // CHECK11-NEXT: [[TMP39:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4 1667 // CHECK11-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP39]]) 1668 // CHECK11-NEXT: ret i32 [[TMP38]] 1669 // 1670 // 1671 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75 1672 // CHECK11-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 1673 // CHECK11-NEXT: entry: 1674 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 1675 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 1676 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 1677 // CHECK11-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 1678 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 1679 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 1680 // CHECK11-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 1681 // CHECK11-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 1682 // CHECK11-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]]) 1683 // CHECK11-NEXT: ret void 1684 // 1685 // 1686 // CHECK11-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z15teams_local_argv_l75.omp_outlined 1687 // CHECK11-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 1688 // CHECK11-NEXT: entry: 1689 // CHECK11-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 1690 // CHECK11-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 1691 // CHECK11-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4 1692 // CHECK11-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 1693 // CHECK11-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 1694 // CHECK11-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1695 // CHECK11-NEXT: [[TMP:%.*]] = alloca i32, align 4 1696 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1697 // CHECK11-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1698 // CHECK11-NEXT: [[I:%.*]] = alloca i32, align 4 1699 // CHECK11-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1700 // CHECK11-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1701 // CHECK11-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 1702 // CHECK11-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 1703 // CHECK11-NEXT: [[I3:%.*]] = alloca i32, align 4 1704 // CHECK11-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 1705 // CHECK11-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 1706 // CHECK11-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4 1707 // CHECK11-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 1708 // CHECK11-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 1709 // CHECK11-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4 1710 // CHECK11-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 1711 // CHECK11-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4 1712 // CHECK11-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 1713 // CHECK11-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 1714 // CHECK11-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1715 // CHECK11-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1716 // CHECK11-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1717 // CHECK11-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1718 // CHECK11-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1719 // CHECK11-NEXT: store i32 0, ptr [[I]], align 4 1720 // CHECK11-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1721 // CHECK11-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1722 // CHECK11-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 1723 // CHECK11: omp.precond.then: 1724 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1725 // CHECK11-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1726 // CHECK11-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 1727 // CHECK11-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 1728 // CHECK11-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 1729 // CHECK11-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1730 // CHECK11-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 1731 // CHECK11-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 1732 // CHECK11-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1733 // CHECK11-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1734 // CHECK11-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 1735 // CHECK11-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 1736 // CHECK11: cond.true: 1737 // CHECK11-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1738 // CHECK11-NEXT: br label [[COND_END:%.*]] 1739 // CHECK11: cond.false: 1740 // CHECK11-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 1741 // CHECK11-NEXT: br label [[COND_END]] 1742 // CHECK11: cond.end: 1743 // CHECK11-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 1744 // CHECK11-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 1745 // CHECK11-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1746 // CHECK11-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 1747 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1748 // CHECK11: omp.inner.for.cond: 1749 // CHECK11-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5:![0-9]+]] 1750 // CHECK11-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP5]] 1751 // CHECK11-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 1752 // CHECK11-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1753 // CHECK11: omp.inner.for.body: 1754 // CHECK11-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 1755 // CHECK11-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 1756 // CHECK11-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1757 // CHECK11-NEXT: store i32 [[ADD]], ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP5]] 1758 // CHECK11-NEXT: [[TMP17:%.*]] = load i32, ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP5]] 1759 // CHECK11-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP17]] 1760 // CHECK11-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP5]] 1761 // CHECK11-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1762 // CHECK11: omp.body.continue: 1763 // CHECK11-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1764 // CHECK11: omp.inner.for.inc: 1765 // CHECK11-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 1766 // CHECK11-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], 1 1767 // CHECK11-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 1768 // CHECK11-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] 1769 // CHECK11: omp.inner.for.end: 1770 // CHECK11-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 1771 // CHECK11: omp.loop.exit: 1772 // CHECK11-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 1773 // CHECK11-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4 1774 // CHECK11-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP20]]) 1775 // CHECK11-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 1776 // CHECK11-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 1777 // CHECK11-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 1778 // CHECK11: .omp.final.then: 1779 // CHECK11-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1780 // CHECK11-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP23]], 0 1781 // CHECK11-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 1782 // CHECK11-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1 1783 // CHECK11-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]] 1784 // CHECK11-NEXT: store i32 [[ADD10]], ptr [[I3]], align 4 1785 // CHECK11-NEXT: br label [[DOTOMP_FINAL_DONE]] 1786 // CHECK11: .omp.final.done: 1787 // CHECK11-NEXT: br label [[OMP_PRECOND_END]] 1788 // CHECK11: omp.precond.end: 1789 // CHECK11-NEXT: ret void 1790 // 1791 // 1792 // CHECK13-LABEL: define {{[^@]+}}@_Z15teams_local_argv 1793 // CHECK13-SAME: () #[[ATTR0:[0-9]+]] { 1794 // CHECK13-NEXT: entry: 1795 // CHECK13-NEXT: [[N:%.*]] = alloca i32, align 4 1796 // CHECK13-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 8 1797 // CHECK13-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 1798 // CHECK13-NEXT: [[TMP:%.*]] = alloca i32, align 4 1799 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1800 // CHECK13-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1801 // CHECK13-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1802 // CHECK13-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1803 // CHECK13-NEXT: [[I:%.*]] = alloca i32, align 4 1804 // CHECK13-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1805 // CHECK13-NEXT: [[I3:%.*]] = alloca i32, align 4 1806 // CHECK13-NEXT: store i32 100, ptr [[N]], align 4 1807 // CHECK13-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 1808 // CHECK13-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 1809 // CHECK13-NEXT: [[TMP2:%.*]] = call ptr @llvm.stacksave.p0() 1810 // CHECK13-NEXT: store ptr [[TMP2]], ptr [[SAVED_STACK]], align 8 1811 // CHECK13-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 1812 // CHECK13-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8 1813 // CHECK13-NEXT: [[TMP3:%.*]] = load i32, ptr [[N]], align 4 1814 // CHECK13-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 1815 // CHECK13-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1816 // CHECK13-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 1817 // CHECK13-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1818 // CHECK13-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1819 // CHECK13-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1820 // CHECK13-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1821 // CHECK13-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1822 // CHECK13-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_UB]], align 4 1823 // CHECK13-NEXT: store i32 0, ptr [[I]], align 4 1824 // CHECK13-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1825 // CHECK13-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 1826 // CHECK13-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 1827 // CHECK13: simd.if.then: 1828 // CHECK13-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1829 // CHECK13-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 1830 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1831 // CHECK13: omp.inner.for.cond: 1832 // CHECK13-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2:![0-9]+]] 1833 // CHECK13-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP2]] 1834 // CHECK13-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 1835 // CHECK13-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1836 // CHECK13: omp.inner.for.body: 1837 // CHECK13-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 1838 // CHECK13-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 1839 // CHECK13-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1840 // CHECK13-NEXT: store i32 [[ADD]], ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP2]] 1841 // CHECK13-NEXT: [[TMP11:%.*]] = load i32, ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP2]] 1842 // CHECK13-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 1843 // CHECK13-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i64 [[IDXPROM]] 1844 // CHECK13-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP2]] 1845 // CHECK13-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1846 // CHECK13: omp.body.continue: 1847 // CHECK13-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1848 // CHECK13: omp.inner.for.inc: 1849 // CHECK13-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 1850 // CHECK13-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP12]], 1 1851 // CHECK13-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 1852 // CHECK13-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] 1853 // CHECK13: omp.inner.for.end: 1854 // CHECK13-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1855 // CHECK13-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP13]], 0 1856 // CHECK13-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 1857 // CHECK13-NEXT: [[MUL8:%.*]] = mul nsw i32 [[DIV7]], 1 1858 // CHECK13-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL8]] 1859 // CHECK13-NEXT: store i32 [[ADD9]], ptr [[I3]], align 4 1860 // CHECK13-NEXT: br label [[SIMD_IF_END]] 1861 // CHECK13: simd.if.end: 1862 // CHECK13-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i64 0 1863 // CHECK13-NEXT: [[TMP14:%.*]] = load i32, ptr [[ARRAYIDX10]], align 4 1864 // CHECK13-NEXT: [[TMP15:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 1865 // CHECK13-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP15]]) 1866 // CHECK13-NEXT: ret i32 [[TMP14]] 1867 // 1868 // 1869 // CHECK15-LABEL: define {{[^@]+}}@_Z15teams_local_argv 1870 // CHECK15-SAME: () #[[ATTR0:[0-9]+]] { 1871 // CHECK15-NEXT: entry: 1872 // CHECK15-NEXT: [[N:%.*]] = alloca i32, align 4 1873 // CHECK15-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 4 1874 // CHECK15-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 1875 // CHECK15-NEXT: [[TMP:%.*]] = alloca i32, align 4 1876 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 1877 // CHECK15-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 1878 // CHECK15-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 1879 // CHECK15-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 1880 // CHECK15-NEXT: [[I:%.*]] = alloca i32, align 4 1881 // CHECK15-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 1882 // CHECK15-NEXT: [[I3:%.*]] = alloca i32, align 4 1883 // CHECK15-NEXT: store i32 100, ptr [[N]], align 4 1884 // CHECK15-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 1885 // CHECK15-NEXT: [[TMP1:%.*]] = call ptr @llvm.stacksave.p0() 1886 // CHECK15-NEXT: store ptr [[TMP1]], ptr [[SAVED_STACK]], align 4 1887 // CHECK15-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 1888 // CHECK15-NEXT: store i32 [[TMP0]], ptr [[__VLA_EXPR0]], align 4 1889 // CHECK15-NEXT: [[TMP2:%.*]] = load i32, ptr [[N]], align 4 1890 // CHECK15-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 1891 // CHECK15-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1892 // CHECK15-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 1893 // CHECK15-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 1894 // CHECK15-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 1895 // CHECK15-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 1896 // CHECK15-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 1897 // CHECK15-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 1898 // CHECK15-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_UB]], align 4 1899 // CHECK15-NEXT: store i32 0, ptr [[I]], align 4 1900 // CHECK15-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1901 // CHECK15-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 1902 // CHECK15-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 1903 // CHECK15: simd.if.then: 1904 // CHECK15-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 1905 // CHECK15-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4 1906 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 1907 // CHECK15: omp.inner.for.cond: 1908 // CHECK15-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3:![0-9]+]] 1909 // CHECK15-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP3]] 1910 // CHECK15-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] 1911 // CHECK15-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 1912 // CHECK15: omp.inner.for.body: 1913 // CHECK15-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 1914 // CHECK15-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 1915 // CHECK15-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 1916 // CHECK15-NEXT: store i32 [[ADD]], ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP3]] 1917 // CHECK15-NEXT: [[TMP10:%.*]] = load i32, ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP3]] 1918 // CHECK15-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i32 [[TMP10]] 1919 // CHECK15-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP3]] 1920 // CHECK15-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 1921 // CHECK15: omp.body.continue: 1922 // CHECK15-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 1923 // CHECK15: omp.inner.for.inc: 1924 // CHECK15-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 1925 // CHECK15-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP11]], 1 1926 // CHECK15-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 1927 // CHECK15-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] 1928 // CHECK15: omp.inner.for.end: 1929 // CHECK15-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 1930 // CHECK15-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP12]], 0 1931 // CHECK15-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 1932 // CHECK15-NEXT: [[MUL8:%.*]] = mul nsw i32 [[DIV7]], 1 1933 // CHECK15-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL8]] 1934 // CHECK15-NEXT: store i32 [[ADD9]], ptr [[I3]], align 4 1935 // CHECK15-NEXT: br label [[SIMD_IF_END]] 1936 // CHECK15: simd.if.end: 1937 // CHECK15-NEXT: [[ARRAYIDX10:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i32 0 1938 // CHECK15-NEXT: [[TMP13:%.*]] = load i32, ptr [[ARRAYIDX10]], align 4 1939 // CHECK15-NEXT: [[TMP14:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4 1940 // CHECK15-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP14]]) 1941 // CHECK15-NEXT: ret i32 [[TMP13]] 1942 // 1943 // 1944 // CHECK17-LABEL: define {{[^@]+}}@_Z21teams_template_structv 1945 // CHECK17-SAME: () #[[ATTR0:[0-9]+]] { 1946 // CHECK17-NEXT: entry: 1947 // CHECK17-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 1948 // CHECK17-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]]) 1949 // CHECK17-NEXT: ret i32 [[CALL]] 1950 // 1951 // 1952 // CHECK17-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 1953 // CHECK17-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat { 1954 // CHECK17-NEXT: entry: 1955 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 1956 // CHECK17-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8 1957 // CHECK17-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8 1958 // CHECK17-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8 1959 // CHECK17-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8 1960 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4 1961 // CHECK17-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 1962 // CHECK17-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 1963 // CHECK17-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 1964 // CHECK17-NEXT: [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 1 1965 // CHECK17-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 1966 // CHECK17-NEXT: [[TMP0:%.*]] = getelementptr float, ptr [[B]], i32 1 1967 // CHECK17-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 1968 // CHECK17-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[A]] to i64 1969 // CHECK17-NEXT: [[TMP3:%.*]] = sub i64 [[TMP1]], [[TMP2]] 1970 // CHECK17-NEXT: [[TMP4:%.*]] = sdiv exact i64 [[TMP3]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) 1971 // CHECK17-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 24, i1 false) 1972 // CHECK17-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1973 // CHECK17-NEXT: store ptr [[THIS1]], ptr [[TMP5]], align 8 1974 // CHECK17-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1975 // CHECK17-NEXT: store ptr [[A]], ptr [[TMP6]], align 8 1976 // CHECK17-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 1977 // CHECK17-NEXT: store i64 [[TMP4]], ptr [[TMP7]], align 8 1978 // CHECK17-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 1979 // CHECK17-NEXT: store ptr null, ptr [[TMP8]], align 8 1980 // CHECK17-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 1981 // CHECK17-NEXT: store ptr [[THIS1]], ptr [[TMP9]], align 8 1982 // CHECK17-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 1983 // CHECK17-NEXT: store ptr [[B]], ptr [[TMP10]], align 8 1984 // CHECK17-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 1985 // CHECK17-NEXT: store ptr null, ptr [[TMP11]], align 8 1986 // CHECK17-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 1987 // CHECK17-NEXT: store ptr [[THIS1]], ptr [[TMP12]], align 8 1988 // CHECK17-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 1989 // CHECK17-NEXT: store ptr [[A]], ptr [[TMP13]], align 8 1990 // CHECK17-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 1991 // CHECK17-NEXT: store ptr null, ptr [[TMP14]], align 8 1992 // CHECK17-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 1993 // CHECK17-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 1994 // CHECK17-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 1995 // CHECK17-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 1996 // CHECK17-NEXT: store i32 3, ptr [[TMP18]], align 4 1997 // CHECK17-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 1998 // CHECK17-NEXT: store i32 3, ptr [[TMP19]], align 4 1999 // CHECK17-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 2000 // CHECK17-NEXT: store ptr [[TMP15]], ptr [[TMP20]], align 8 2001 // CHECK17-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 2002 // CHECK17-NEXT: store ptr [[TMP16]], ptr [[TMP21]], align 8 2003 // CHECK17-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 2004 // CHECK17-NEXT: store ptr [[TMP17]], ptr [[TMP22]], align 8 2005 // CHECK17-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 2006 // CHECK17-NEXT: store ptr @.offload_maptypes, ptr [[TMP23]], align 8 2007 // CHECK17-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 2008 // CHECK17-NEXT: store ptr null, ptr [[TMP24]], align 8 2009 // CHECK17-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 2010 // CHECK17-NEXT: store ptr null, ptr [[TMP25]], align 8 2011 // CHECK17-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 2012 // CHECK17-NEXT: store i64 123, ptr [[TMP26]], align 8 2013 // CHECK17-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 2014 // CHECK17-NEXT: store i64 0, ptr [[TMP27]], align 8 2015 // CHECK17-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 2016 // CHECK17-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP28]], align 4 2017 // CHECK17-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 2018 // CHECK17-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP29]], align 4 2019 // CHECK17-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 2020 // CHECK17-NEXT: store i32 0, ptr [[TMP30]], align 4 2021 // CHECK17-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, ptr [[KERNEL_ARGS]]) 2022 // CHECK17-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 2023 // CHECK17-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2024 // CHECK17: omp_offload.failed: 2025 // CHECK17-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(ptr [[THIS1]]) #[[ATTR2:[0-9]+]] 2026 // CHECK17-NEXT: br label [[OMP_OFFLOAD_CONT]] 2027 // CHECK17: omp_offload.cont: 2028 // CHECK17-NEXT: [[A2:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 2029 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A2]], i64 0, i64 0 2030 // CHECK17-NEXT: [[TMP33:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 2031 // CHECK17-NEXT: ret i32 [[TMP33]] 2032 // 2033 // 2034 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123 2035 // CHECK17-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 2036 // CHECK17-NEXT: entry: 2037 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 2038 // CHECK17-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 2039 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 2040 // CHECK17-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.omp_outlined, ptr [[TMP0]]) 2041 // CHECK17-NEXT: ret void 2042 // 2043 // 2044 // CHECK17-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.omp_outlined 2045 // CHECK17-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 2046 // CHECK17-NEXT: entry: 2047 // CHECK17-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 2048 // CHECK17-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 2049 // CHECK17-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 2050 // CHECK17-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2051 // CHECK17-NEXT: [[TMP:%.*]] = alloca i32, align 4 2052 // CHECK17-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2053 // CHECK17-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2054 // CHECK17-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2055 // CHECK17-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2056 // CHECK17-NEXT: [[I:%.*]] = alloca i32, align 4 2057 // CHECK17-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 2058 // CHECK17-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 2059 // CHECK17-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 2060 // CHECK17-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 2061 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2062 // CHECK17-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 2063 // CHECK17-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2064 // CHECK17-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2065 // CHECK17-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2066 // CHECK17-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 2067 // CHECK17-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2068 // CHECK17-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2069 // CHECK17-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 2070 // CHECK17-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2071 // CHECK17: cond.true: 2072 // CHECK17-NEXT: br label [[COND_END:%.*]] 2073 // CHECK17: cond.false: 2074 // CHECK17-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2075 // CHECK17-NEXT: br label [[COND_END]] 2076 // CHECK17: cond.end: 2077 // CHECK17-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2078 // CHECK17-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2079 // CHECK17-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2080 // CHECK17-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 2081 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2082 // CHECK17: omp.inner.for.cond: 2083 // CHECK17-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4:![0-9]+]] 2084 // CHECK17-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP4]] 2085 // CHECK17-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 2086 // CHECK17-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2087 // CHECK17: omp.inner.for.body: 2088 // CHECK17-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]] 2089 // CHECK17-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 2090 // CHECK17-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2091 // CHECK17-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP4]] 2092 // CHECK17-NEXT: [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 1 2093 // CHECK17-NEXT: [[TMP9:%.*]] = load float, ptr [[B]], align 4, !llvm.access.group [[ACC_GRP4]] 2094 // CHECK17-NEXT: [[CONV:%.*]] = fptosi float [[TMP9]] to i32 2095 // CHECK17-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[TMP0]], i32 0, i32 0 2096 // CHECK17-NEXT: [[TMP10:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP4]] 2097 // CHECK17-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP10]] to i64 2098 // CHECK17-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]] 2099 // CHECK17-NEXT: store i32 [[CONV]], ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP4]] 2100 // CHECK17-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2101 // CHECK17: omp.body.continue: 2102 // CHECK17-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2103 // CHECK17: omp.inner.for.inc: 2104 // CHECK17-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]] 2105 // CHECK17-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1 2106 // CHECK17-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]] 2107 // CHECK17-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]] 2108 // CHECK17: omp.inner.for.end: 2109 // CHECK17-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2110 // CHECK17: omp.loop.exit: 2111 // CHECK17-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 2112 // CHECK17-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 2113 // CHECK17-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 2114 // CHECK17-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 2115 // CHECK17: .omp.final.then: 2116 // CHECK17-NEXT: store i32 123, ptr [[I]], align 4 2117 // CHECK17-NEXT: br label [[DOTOMP_FINAL_DONE]] 2118 // CHECK17: .omp.final.done: 2119 // CHECK17-NEXT: ret void 2120 // 2121 // 2122 // CHECK19-LABEL: define {{[^@]+}}@_Z21teams_template_structv 2123 // CHECK19-SAME: () #[[ATTR0:[0-9]+]] { 2124 // CHECK19-NEXT: entry: 2125 // CHECK19-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 2126 // CHECK19-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]]) 2127 // CHECK19-NEXT: ret i32 [[CALL]] 2128 // 2129 // 2130 // CHECK19-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 2131 // CHECK19-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 2132 // CHECK19-NEXT: entry: 2133 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 2134 // CHECK19-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4 2135 // CHECK19-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4 2136 // CHECK19-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4 2137 // CHECK19-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4 2138 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4 2139 // CHECK19-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 2140 // CHECK19-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 2141 // CHECK19-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 2142 // CHECK19-NEXT: [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 1 2143 // CHECK19-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 2144 // CHECK19-NEXT: [[TMP0:%.*]] = getelementptr float, ptr [[B]], i32 1 2145 // CHECK19-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 2146 // CHECK19-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[A]] to i64 2147 // CHECK19-NEXT: [[TMP3:%.*]] = sub i64 [[TMP1]], [[TMP2]] 2148 // CHECK19-NEXT: [[TMP4:%.*]] = sdiv exact i64 [[TMP3]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) 2149 // CHECK19-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 24, i1 false) 2150 // CHECK19-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2151 // CHECK19-NEXT: store ptr [[THIS1]], ptr [[TMP5]], align 4 2152 // CHECK19-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2153 // CHECK19-NEXT: store ptr [[A]], ptr [[TMP6]], align 4 2154 // CHECK19-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 2155 // CHECK19-NEXT: store i64 [[TMP4]], ptr [[TMP7]], align 4 2156 // CHECK19-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 2157 // CHECK19-NEXT: store ptr null, ptr [[TMP8]], align 4 2158 // CHECK19-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 2159 // CHECK19-NEXT: store ptr [[THIS1]], ptr [[TMP9]], align 4 2160 // CHECK19-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 2161 // CHECK19-NEXT: store ptr [[B]], ptr [[TMP10]], align 4 2162 // CHECK19-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 2163 // CHECK19-NEXT: store ptr null, ptr [[TMP11]], align 4 2164 // CHECK19-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 2165 // CHECK19-NEXT: store ptr [[THIS1]], ptr [[TMP12]], align 4 2166 // CHECK19-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 2167 // CHECK19-NEXT: store ptr [[A]], ptr [[TMP13]], align 4 2168 // CHECK19-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 2169 // CHECK19-NEXT: store ptr null, ptr [[TMP14]], align 4 2170 // CHECK19-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2171 // CHECK19-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2172 // CHECK19-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 2173 // CHECK19-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 2174 // CHECK19-NEXT: store i32 3, ptr [[TMP18]], align 4 2175 // CHECK19-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 2176 // CHECK19-NEXT: store i32 3, ptr [[TMP19]], align 4 2177 // CHECK19-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 2178 // CHECK19-NEXT: store ptr [[TMP15]], ptr [[TMP20]], align 4 2179 // CHECK19-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 2180 // CHECK19-NEXT: store ptr [[TMP16]], ptr [[TMP21]], align 4 2181 // CHECK19-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 2182 // CHECK19-NEXT: store ptr [[TMP17]], ptr [[TMP22]], align 4 2183 // CHECK19-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 2184 // CHECK19-NEXT: store ptr @.offload_maptypes, ptr [[TMP23]], align 4 2185 // CHECK19-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 2186 // CHECK19-NEXT: store ptr null, ptr [[TMP24]], align 4 2187 // CHECK19-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 2188 // CHECK19-NEXT: store ptr null, ptr [[TMP25]], align 4 2189 // CHECK19-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 2190 // CHECK19-NEXT: store i64 123, ptr [[TMP26]], align 8 2191 // CHECK19-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 2192 // CHECK19-NEXT: store i64 0, ptr [[TMP27]], align 8 2193 // CHECK19-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 2194 // CHECK19-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP28]], align 4 2195 // CHECK19-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 2196 // CHECK19-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP29]], align 4 2197 // CHECK19-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 2198 // CHECK19-NEXT: store i32 0, ptr [[TMP30]], align 4 2199 // CHECK19-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, ptr [[KERNEL_ARGS]]) 2200 // CHECK19-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 2201 // CHECK19-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2202 // CHECK19: omp_offload.failed: 2203 // CHECK19-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(ptr [[THIS1]]) #[[ATTR2:[0-9]+]] 2204 // CHECK19-NEXT: br label [[OMP_OFFLOAD_CONT]] 2205 // CHECK19: omp_offload.cont: 2206 // CHECK19-NEXT: [[A2:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 2207 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A2]], i32 0, i32 0 2208 // CHECK19-NEXT: [[TMP33:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 2209 // CHECK19-NEXT: ret i32 [[TMP33]] 2210 // 2211 // 2212 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123 2213 // CHECK19-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 2214 // CHECK19-NEXT: entry: 2215 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 2216 // CHECK19-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 2217 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 2218 // CHECK19-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.omp_outlined, ptr [[TMP0]]) 2219 // CHECK19-NEXT: ret void 2220 // 2221 // 2222 // CHECK19-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.omp_outlined 2223 // CHECK19-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]]) #[[ATTR1]] { 2224 // CHECK19-NEXT: entry: 2225 // CHECK19-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 2226 // CHECK19-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 2227 // CHECK19-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 2228 // CHECK19-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2229 // CHECK19-NEXT: [[TMP:%.*]] = alloca i32, align 4 2230 // CHECK19-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2231 // CHECK19-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2232 // CHECK19-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2233 // CHECK19-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2234 // CHECK19-NEXT: [[I:%.*]] = alloca i32, align 4 2235 // CHECK19-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 2236 // CHECK19-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 2237 // CHECK19-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 2238 // CHECK19-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 2239 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2240 // CHECK19-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 2241 // CHECK19-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2242 // CHECK19-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2243 // CHECK19-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 2244 // CHECK19-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 2245 // CHECK19-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2246 // CHECK19-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2247 // CHECK19-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 2248 // CHECK19-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2249 // CHECK19: cond.true: 2250 // CHECK19-NEXT: br label [[COND_END:%.*]] 2251 // CHECK19: cond.false: 2252 // CHECK19-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2253 // CHECK19-NEXT: br label [[COND_END]] 2254 // CHECK19: cond.end: 2255 // CHECK19-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2256 // CHECK19-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2257 // CHECK19-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2258 // CHECK19-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 2259 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2260 // CHECK19: omp.inner.for.cond: 2261 // CHECK19-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5:![0-9]+]] 2262 // CHECK19-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP5]] 2263 // CHECK19-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 2264 // CHECK19-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2265 // CHECK19: omp.inner.for.body: 2266 // CHECK19-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 2267 // CHECK19-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 2268 // CHECK19-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2269 // CHECK19-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP5]] 2270 // CHECK19-NEXT: [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 1 2271 // CHECK19-NEXT: [[TMP9:%.*]] = load float, ptr [[B]], align 4, !llvm.access.group [[ACC_GRP5]] 2272 // CHECK19-NEXT: [[CONV:%.*]] = fptosi float [[TMP9]] to i32 2273 // CHECK19-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[TMP0]], i32 0, i32 0 2274 // CHECK19-NEXT: [[TMP10:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP5]] 2275 // CHECK19-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP10]] 2276 // CHECK19-NEXT: store i32 [[CONV]], ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP5]] 2277 // CHECK19-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2278 // CHECK19: omp.body.continue: 2279 // CHECK19-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2280 // CHECK19: omp.inner.for.inc: 2281 // CHECK19-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 2282 // CHECK19-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP11]], 1 2283 // CHECK19-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 2284 // CHECK19-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] 2285 // CHECK19: omp.inner.for.end: 2286 // CHECK19-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2287 // CHECK19: omp.loop.exit: 2288 // CHECK19-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 2289 // CHECK19-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 2290 // CHECK19-NEXT: [[TMP13:%.*]] = icmp ne i32 [[TMP12]], 0 2291 // CHECK19-NEXT: br i1 [[TMP13]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 2292 // CHECK19: .omp.final.then: 2293 // CHECK19-NEXT: store i32 123, ptr [[I]], align 4 2294 // CHECK19-NEXT: br label [[DOTOMP_FINAL_DONE]] 2295 // CHECK19: .omp.final.done: 2296 // CHECK19-NEXT: ret void 2297 // 2298 // 2299 // CHECK21-LABEL: define {{[^@]+}}@_Z21teams_template_structv 2300 // CHECK21-SAME: () #[[ATTR0:[0-9]+]] { 2301 // CHECK21-NEXT: entry: 2302 // CHECK21-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 2303 // CHECK21-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]]) 2304 // CHECK21-NEXT: ret i32 [[CALL]] 2305 // 2306 // 2307 // CHECK21-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 2308 // CHECK21-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat { 2309 // CHECK21-NEXT: entry: 2310 // CHECK21-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 2311 // CHECK21-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8 2312 // CHECK21-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8 2313 // CHECK21-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8 2314 // CHECK21-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8 2315 // CHECK21-NEXT: [[TMP:%.*]] = alloca i32, align 4 2316 // CHECK21-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 2317 // CHECK21-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 2318 // CHECK21-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 2319 // CHECK21-NEXT: [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 1 2320 // CHECK21-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 2321 // CHECK21-NEXT: [[TMP0:%.*]] = getelementptr float, ptr [[B]], i32 1 2322 // CHECK21-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 2323 // CHECK21-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[A]] to i64 2324 // CHECK21-NEXT: [[TMP3:%.*]] = sub i64 [[TMP1]], [[TMP2]] 2325 // CHECK21-NEXT: [[TMP4:%.*]] = sdiv exact i64 [[TMP3]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) 2326 // CHECK21-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 24, i1 false) 2327 // CHECK21-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2328 // CHECK21-NEXT: store ptr [[THIS1]], ptr [[TMP5]], align 8 2329 // CHECK21-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2330 // CHECK21-NEXT: store ptr [[A]], ptr [[TMP6]], align 8 2331 // CHECK21-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 2332 // CHECK21-NEXT: store i64 [[TMP4]], ptr [[TMP7]], align 8 2333 // CHECK21-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 2334 // CHECK21-NEXT: store ptr null, ptr [[TMP8]], align 8 2335 // CHECK21-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 2336 // CHECK21-NEXT: store ptr [[THIS1]], ptr [[TMP9]], align 8 2337 // CHECK21-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 2338 // CHECK21-NEXT: store ptr [[B]], ptr [[TMP10]], align 8 2339 // CHECK21-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 2340 // CHECK21-NEXT: store ptr null, ptr [[TMP11]], align 8 2341 // CHECK21-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 2342 // CHECK21-NEXT: store ptr [[THIS1]], ptr [[TMP12]], align 8 2343 // CHECK21-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 2344 // CHECK21-NEXT: store ptr [[A]], ptr [[TMP13]], align 8 2345 // CHECK21-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 2346 // CHECK21-NEXT: store ptr null, ptr [[TMP14]], align 8 2347 // CHECK21-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2348 // CHECK21-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2349 // CHECK21-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 2350 // CHECK21-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 2351 // CHECK21-NEXT: store i32 3, ptr [[TMP18]], align 4 2352 // CHECK21-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 2353 // CHECK21-NEXT: store i32 3, ptr [[TMP19]], align 4 2354 // CHECK21-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 2355 // CHECK21-NEXT: store ptr [[TMP15]], ptr [[TMP20]], align 8 2356 // CHECK21-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 2357 // CHECK21-NEXT: store ptr [[TMP16]], ptr [[TMP21]], align 8 2358 // CHECK21-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 2359 // CHECK21-NEXT: store ptr [[TMP17]], ptr [[TMP22]], align 8 2360 // CHECK21-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 2361 // CHECK21-NEXT: store ptr @.offload_maptypes, ptr [[TMP23]], align 8 2362 // CHECK21-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 2363 // CHECK21-NEXT: store ptr null, ptr [[TMP24]], align 8 2364 // CHECK21-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 2365 // CHECK21-NEXT: store ptr null, ptr [[TMP25]], align 8 2366 // CHECK21-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 2367 // CHECK21-NEXT: store i64 123, ptr [[TMP26]], align 8 2368 // CHECK21-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 2369 // CHECK21-NEXT: store i64 0, ptr [[TMP27]], align 8 2370 // CHECK21-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 2371 // CHECK21-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP28]], align 4 2372 // CHECK21-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 2373 // CHECK21-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP29]], align 4 2374 // CHECK21-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 2375 // CHECK21-NEXT: store i32 0, ptr [[TMP30]], align 4 2376 // CHECK21-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, ptr [[KERNEL_ARGS]]) 2377 // CHECK21-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 2378 // CHECK21-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2379 // CHECK21: omp_offload.failed: 2380 // CHECK21-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(ptr [[THIS1]]) #[[ATTR2:[0-9]+]] 2381 // CHECK21-NEXT: br label [[OMP_OFFLOAD_CONT]] 2382 // CHECK21: omp_offload.cont: 2383 // CHECK21-NEXT: [[A2:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 2384 // CHECK21-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A2]], i64 0, i64 0 2385 // CHECK21-NEXT: [[TMP33:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 2386 // CHECK21-NEXT: ret i32 [[TMP33]] 2387 // 2388 // 2389 // CHECK21-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123 2390 // CHECK21-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 2391 // CHECK21-NEXT: entry: 2392 // CHECK21-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 2393 // CHECK21-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 2394 // CHECK21-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 2395 // CHECK21-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 2396 // CHECK21-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 2397 // CHECK21-NEXT: [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 1 2398 // CHECK21-NEXT: [[TMP1:%.*]] = load float, ptr [[B]], align 4 2399 // CHECK21-NEXT: [[TOBOOL:%.*]] = fcmp une float [[TMP1]], 0.000000e+00 2400 // CHECK21-NEXT: [[STOREDV:%.*]] = zext i1 [[TOBOOL]] to i8 2401 // CHECK21-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR_]], align 1 2402 // CHECK21-NEXT: [[TMP2:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1 2403 // CHECK21-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP2]] to i1 2404 // CHECK21-NEXT: [[STOREDV1:%.*]] = zext i1 [[LOADEDV]] to i8 2405 // CHECK21-NEXT: store i8 [[STOREDV1]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1 2406 // CHECK21-NEXT: [[TMP3:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8 2407 // CHECK21-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.omp_outlined, ptr [[TMP0]], i64 [[TMP3]]) 2408 // CHECK21-NEXT: ret void 2409 // 2410 // 2411 // CHECK21-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.omp_outlined 2412 // CHECK21-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] { 2413 // CHECK21-NEXT: entry: 2414 // CHECK21-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 2415 // CHECK21-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 2416 // CHECK21-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 2417 // CHECK21-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 2418 // CHECK21-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2419 // CHECK21-NEXT: [[TMP:%.*]] = alloca i32, align 4 2420 // CHECK21-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2421 // CHECK21-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2422 // CHECK21-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2423 // CHECK21-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2424 // CHECK21-NEXT: [[I:%.*]] = alloca i32, align 4 2425 // CHECK21-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 2426 // CHECK21-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 2427 // CHECK21-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 2428 // CHECK21-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8 2429 // CHECK21-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 2430 // CHECK21-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2431 // CHECK21-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 2432 // CHECK21-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2433 // CHECK21-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2434 // CHECK21-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 2435 // CHECK21-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 2436 // CHECK21-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2437 // CHECK21-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2438 // CHECK21-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 2439 // CHECK21-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2440 // CHECK21: cond.true: 2441 // CHECK21-NEXT: br label [[COND_END:%.*]] 2442 // CHECK21: cond.false: 2443 // CHECK21-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2444 // CHECK21-NEXT: br label [[COND_END]] 2445 // CHECK21: cond.end: 2446 // CHECK21-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2447 // CHECK21-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2448 // CHECK21-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2449 // CHECK21-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 2450 // CHECK21-NEXT: [[TMP6:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1 2451 // CHECK21-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 2452 // CHECK21-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] 2453 // CHECK21: omp_if.then: 2454 // CHECK21-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2455 // CHECK21: omp.inner.for.cond: 2456 // CHECK21-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4:![0-9]+]] 2457 // CHECK21-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP4]] 2458 // CHECK21-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] 2459 // CHECK21-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2460 // CHECK21: omp.inner.for.body: 2461 // CHECK21-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]] 2462 // CHECK21-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 2463 // CHECK21-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2464 // CHECK21-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP4]] 2465 // CHECK21-NEXT: [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 1 2466 // CHECK21-NEXT: [[TMP10:%.*]] = load float, ptr [[B]], align 4, !nontemporal [[META5:![0-9]+]], !llvm.access.group [[ACC_GRP4]] 2467 // CHECK21-NEXT: [[CONV:%.*]] = fptosi float [[TMP10]] to i32 2468 // CHECK21-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[TMP0]], i32 0, i32 0 2469 // CHECK21-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP4]] 2470 // CHECK21-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 2471 // CHECK21-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]] 2472 // CHECK21-NEXT: store i32 [[CONV]], ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP4]] 2473 // CHECK21-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2474 // CHECK21: omp.body.continue: 2475 // CHECK21-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2476 // CHECK21: omp.inner.for.inc: 2477 // CHECK21-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]] 2478 // CHECK21-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1 2479 // CHECK21-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP4]] 2480 // CHECK21-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] 2481 // CHECK21: omp.inner.for.end: 2482 // CHECK21-NEXT: br label [[OMP_IF_END:%.*]] 2483 // CHECK21: omp_if.else: 2484 // CHECK21-NEXT: br label [[OMP_INNER_FOR_COND3:%.*]] 2485 // CHECK21: omp.inner.for.cond3: 2486 // CHECK21-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2487 // CHECK21-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2488 // CHECK21-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 2489 // CHECK21-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY5:%.*]], label [[OMP_INNER_FOR_END16:%.*]] 2490 // CHECK21: omp.inner.for.body5: 2491 // CHECK21-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2492 // CHECK21-NEXT: [[MUL6:%.*]] = mul nsw i32 [[TMP15]], 1 2493 // CHECK21-NEXT: [[ADD7:%.*]] = add nsw i32 0, [[MUL6]] 2494 // CHECK21-NEXT: store i32 [[ADD7]], ptr [[I]], align 4 2495 // CHECK21-NEXT: [[B8:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[TMP0]], i32 0, i32 1 2496 // CHECK21-NEXT: [[TMP16:%.*]] = load float, ptr [[B8]], align 4 2497 // CHECK21-NEXT: [[CONV9:%.*]] = fptosi float [[TMP16]] to i32 2498 // CHECK21-NEXT: [[A10:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[TMP0]], i32 0, i32 0 2499 // CHECK21-NEXT: [[TMP17:%.*]] = load i32, ptr [[I]], align 4 2500 // CHECK21-NEXT: [[IDXPROM11:%.*]] = sext i32 [[TMP17]] to i64 2501 // CHECK21-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds [123 x i32], ptr [[A10]], i64 0, i64 [[IDXPROM11]] 2502 // CHECK21-NEXT: store i32 [[CONV9]], ptr [[ARRAYIDX12]], align 4 2503 // CHECK21-NEXT: br label [[OMP_BODY_CONTINUE13:%.*]] 2504 // CHECK21: omp.body.continue13: 2505 // CHECK21-NEXT: br label [[OMP_INNER_FOR_INC14:%.*]] 2506 // CHECK21: omp.inner.for.inc14: 2507 // CHECK21-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2508 // CHECK21-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP18]], 1 2509 // CHECK21-NEXT: store i32 [[ADD15]], ptr [[DOTOMP_IV]], align 4 2510 // CHECK21-NEXT: br label [[OMP_INNER_FOR_COND3]], !llvm.loop [[LOOP9:![0-9]+]] 2511 // CHECK21: omp.inner.for.end16: 2512 // CHECK21-NEXT: br label [[OMP_IF_END]] 2513 // CHECK21: omp_if.end: 2514 // CHECK21-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2515 // CHECK21: omp.loop.exit: 2516 // CHECK21-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 2517 // CHECK21-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 2518 // CHECK21-NEXT: [[TMP20:%.*]] = icmp ne i32 [[TMP19]], 0 2519 // CHECK21-NEXT: br i1 [[TMP20]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 2520 // CHECK21: .omp.final.then: 2521 // CHECK21-NEXT: store i32 123, ptr [[I]], align 4 2522 // CHECK21-NEXT: br label [[DOTOMP_FINAL_DONE]] 2523 // CHECK21: .omp.final.done: 2524 // CHECK21-NEXT: ret void 2525 // 2526 // 2527 // CHECK23-LABEL: define {{[^@]+}}@_Z21teams_template_structv 2528 // CHECK23-SAME: () #[[ATTR0:[0-9]+]] { 2529 // CHECK23-NEXT: entry: 2530 // CHECK23-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 2531 // CHECK23-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]]) 2532 // CHECK23-NEXT: ret i32 [[CALL]] 2533 // 2534 // 2535 // CHECK23-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 2536 // CHECK23-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 2537 // CHECK23-NEXT: entry: 2538 // CHECK23-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 2539 // CHECK23-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4 2540 // CHECK23-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4 2541 // CHECK23-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4 2542 // CHECK23-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4 2543 // CHECK23-NEXT: [[TMP:%.*]] = alloca i32, align 4 2544 // CHECK23-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 2545 // CHECK23-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 2546 // CHECK23-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 2547 // CHECK23-NEXT: [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 1 2548 // CHECK23-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 2549 // CHECK23-NEXT: [[TMP0:%.*]] = getelementptr float, ptr [[B]], i32 1 2550 // CHECK23-NEXT: [[TMP1:%.*]] = ptrtoint ptr [[TMP0]] to i64 2551 // CHECK23-NEXT: [[TMP2:%.*]] = ptrtoint ptr [[A]] to i64 2552 // CHECK23-NEXT: [[TMP3:%.*]] = sub i64 [[TMP1]], [[TMP2]] 2553 // CHECK23-NEXT: [[TMP4:%.*]] = sdiv exact i64 [[TMP3]], ptrtoint (ptr getelementptr (i8, ptr null, i32 1) to i64) 2554 // CHECK23-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 24, i1 false) 2555 // CHECK23-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2556 // CHECK23-NEXT: store ptr [[THIS1]], ptr [[TMP5]], align 4 2557 // CHECK23-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2558 // CHECK23-NEXT: store ptr [[A]], ptr [[TMP6]], align 4 2559 // CHECK23-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 2560 // CHECK23-NEXT: store i64 [[TMP4]], ptr [[TMP7]], align 4 2561 // CHECK23-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 2562 // CHECK23-NEXT: store ptr null, ptr [[TMP8]], align 4 2563 // CHECK23-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 2564 // CHECK23-NEXT: store ptr [[THIS1]], ptr [[TMP9]], align 4 2565 // CHECK23-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 2566 // CHECK23-NEXT: store ptr [[B]], ptr [[TMP10]], align 4 2567 // CHECK23-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 2568 // CHECK23-NEXT: store ptr null, ptr [[TMP11]], align 4 2569 // CHECK23-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 2570 // CHECK23-NEXT: store ptr [[THIS1]], ptr [[TMP12]], align 4 2571 // CHECK23-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 2572 // CHECK23-NEXT: store ptr [[A]], ptr [[TMP13]], align 4 2573 // CHECK23-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 2574 // CHECK23-NEXT: store ptr null, ptr [[TMP14]], align 4 2575 // CHECK23-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 2576 // CHECK23-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 2577 // CHECK23-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 2578 // CHECK23-NEXT: [[TMP18:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 2579 // CHECK23-NEXT: store i32 3, ptr [[TMP18]], align 4 2580 // CHECK23-NEXT: [[TMP19:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 2581 // CHECK23-NEXT: store i32 3, ptr [[TMP19]], align 4 2582 // CHECK23-NEXT: [[TMP20:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 2583 // CHECK23-NEXT: store ptr [[TMP15]], ptr [[TMP20]], align 4 2584 // CHECK23-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 2585 // CHECK23-NEXT: store ptr [[TMP16]], ptr [[TMP21]], align 4 2586 // CHECK23-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 2587 // CHECK23-NEXT: store ptr [[TMP17]], ptr [[TMP22]], align 4 2588 // CHECK23-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 2589 // CHECK23-NEXT: store ptr @.offload_maptypes, ptr [[TMP23]], align 4 2590 // CHECK23-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 2591 // CHECK23-NEXT: store ptr null, ptr [[TMP24]], align 4 2592 // CHECK23-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 2593 // CHECK23-NEXT: store ptr null, ptr [[TMP25]], align 4 2594 // CHECK23-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 2595 // CHECK23-NEXT: store i64 123, ptr [[TMP26]], align 8 2596 // CHECK23-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 2597 // CHECK23-NEXT: store i64 0, ptr [[TMP27]], align 8 2598 // CHECK23-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 2599 // CHECK23-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP28]], align 4 2600 // CHECK23-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 2601 // CHECK23-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP29]], align 4 2602 // CHECK23-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 2603 // CHECK23-NEXT: store i32 0, ptr [[TMP30]], align 4 2604 // CHECK23-NEXT: [[TMP31:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.region_id, ptr [[KERNEL_ARGS]]) 2605 // CHECK23-NEXT: [[TMP32:%.*]] = icmp ne i32 [[TMP31]], 0 2606 // CHECK23-NEXT: br i1 [[TMP32]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 2607 // CHECK23: omp_offload.failed: 2608 // CHECK23-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123(ptr [[THIS1]]) #[[ATTR2:[0-9]+]] 2609 // CHECK23-NEXT: br label [[OMP_OFFLOAD_CONT]] 2610 // CHECK23: omp_offload.cont: 2611 // CHECK23-NEXT: [[A2:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 2612 // CHECK23-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A2]], i32 0, i32 0 2613 // CHECK23-NEXT: [[TMP33:%.*]] = load i32, ptr [[ARRAYIDX]], align 4 2614 // CHECK23-NEXT: ret i32 [[TMP33]] 2615 // 2616 // 2617 // CHECK23-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123 2618 // CHECK23-SAME: (ptr noundef [[THIS:%.*]]) #[[ATTR1:[0-9]+]] { 2619 // CHECK23-NEXT: entry: 2620 // CHECK23-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 2621 // CHECK23-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 2622 // CHECK23-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 2623 // CHECK23-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 2624 // CHECK23-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 2625 // CHECK23-NEXT: [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 1 2626 // CHECK23-NEXT: [[TMP1:%.*]] = load float, ptr [[B]], align 4 2627 // CHECK23-NEXT: [[TOBOOL:%.*]] = fcmp une float [[TMP1]], 0.000000e+00 2628 // CHECK23-NEXT: [[STOREDV:%.*]] = zext i1 [[TOBOOL]] to i8 2629 // CHECK23-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR_]], align 1 2630 // CHECK23-NEXT: [[TMP2:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1 2631 // CHECK23-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP2]] to i1 2632 // CHECK23-NEXT: [[STOREDV1:%.*]] = zext i1 [[LOADEDV]] to i8 2633 // CHECK23-NEXT: store i8 [[STOREDV1]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1 2634 // CHECK23-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 2635 // CHECK23-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 2, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.omp_outlined, ptr [[TMP0]], i32 [[TMP3]]) 2636 // CHECK23-NEXT: ret void 2637 // 2638 // 2639 // CHECK23-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__ZN2SSIiLi123ELx456EE3fooEv_l123.omp_outlined 2640 // CHECK23-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef [[THIS:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR1]] { 2641 // CHECK23-NEXT: entry: 2642 // CHECK23-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 2643 // CHECK23-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 2644 // CHECK23-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 2645 // CHECK23-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 2646 // CHECK23-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2647 // CHECK23-NEXT: [[TMP:%.*]] = alloca i32, align 4 2648 // CHECK23-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2649 // CHECK23-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2650 // CHECK23-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 2651 // CHECK23-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 2652 // CHECK23-NEXT: [[I:%.*]] = alloca i32, align 4 2653 // CHECK23-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 2654 // CHECK23-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 2655 // CHECK23-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 2656 // CHECK23-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 2657 // CHECK23-NEXT: [[TMP0:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 2658 // CHECK23-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2659 // CHECK23-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 2660 // CHECK23-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 2661 // CHECK23-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 2662 // CHECK23-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 2663 // CHECK23-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 2664 // CHECK23-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 2665 // CHECK23-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2666 // CHECK23-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 122 2667 // CHECK23-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 2668 // CHECK23: cond.true: 2669 // CHECK23-NEXT: br label [[COND_END:%.*]] 2670 // CHECK23: cond.false: 2671 // CHECK23-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2672 // CHECK23-NEXT: br label [[COND_END]] 2673 // CHECK23: cond.end: 2674 // CHECK23-NEXT: [[COND:%.*]] = phi i32 [ 122, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 2675 // CHECK23-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 2676 // CHECK23-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2677 // CHECK23-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 2678 // CHECK23-NEXT: [[TMP6:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1 2679 // CHECK23-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP6]] to i1 2680 // CHECK23-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] 2681 // CHECK23: omp_if.then: 2682 // CHECK23-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2683 // CHECK23: omp.inner.for.cond: 2684 // CHECK23-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5:![0-9]+]] 2685 // CHECK23-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP5]] 2686 // CHECK23-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] 2687 // CHECK23-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2688 // CHECK23: omp.inner.for.body: 2689 // CHECK23-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 2690 // CHECK23-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 2691 // CHECK23-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2692 // CHECK23-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP5]] 2693 // CHECK23-NEXT: [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[TMP0]], i32 0, i32 1 2694 // CHECK23-NEXT: [[TMP10:%.*]] = load float, ptr [[B]], align 4, !nontemporal [[META6:![0-9]+]], !llvm.access.group [[ACC_GRP5]] 2695 // CHECK23-NEXT: [[CONV:%.*]] = fptosi float [[TMP10]] to i32 2696 // CHECK23-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[TMP0]], i32 0, i32 0 2697 // CHECK23-NEXT: [[TMP11:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP5]] 2698 // CHECK23-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP11]] 2699 // CHECK23-NEXT: store i32 [[CONV]], ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP5]] 2700 // CHECK23-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2701 // CHECK23: omp.body.continue: 2702 // CHECK23-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2703 // CHECK23: omp.inner.for.inc: 2704 // CHECK23-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 2705 // CHECK23-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP12]], 1 2706 // CHECK23-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 2707 // CHECK23-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] 2708 // CHECK23: omp.inner.for.end: 2709 // CHECK23-NEXT: br label [[OMP_IF_END:%.*]] 2710 // CHECK23: omp_if.else: 2711 // CHECK23-NEXT: br label [[OMP_INNER_FOR_COND3:%.*]] 2712 // CHECK23: omp.inner.for.cond3: 2713 // CHECK23-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2714 // CHECK23-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2715 // CHECK23-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]] 2716 // CHECK23-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY5:%.*]], label [[OMP_INNER_FOR_END15:%.*]] 2717 // CHECK23: omp.inner.for.body5: 2718 // CHECK23-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2719 // CHECK23-NEXT: [[MUL6:%.*]] = mul nsw i32 [[TMP15]], 1 2720 // CHECK23-NEXT: [[ADD7:%.*]] = add nsw i32 0, [[MUL6]] 2721 // CHECK23-NEXT: store i32 [[ADD7]], ptr [[I]], align 4 2722 // CHECK23-NEXT: [[B8:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[TMP0]], i32 0, i32 1 2723 // CHECK23-NEXT: [[TMP16:%.*]] = load float, ptr [[B8]], align 4 2724 // CHECK23-NEXT: [[CONV9:%.*]] = fptosi float [[TMP16]] to i32 2725 // CHECK23-NEXT: [[A10:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[TMP0]], i32 0, i32 0 2726 // CHECK23-NEXT: [[TMP17:%.*]] = load i32, ptr [[I]], align 4 2727 // CHECK23-NEXT: [[ARRAYIDX11:%.*]] = getelementptr inbounds [123 x i32], ptr [[A10]], i32 0, i32 [[TMP17]] 2728 // CHECK23-NEXT: store i32 [[CONV9]], ptr [[ARRAYIDX11]], align 4 2729 // CHECK23-NEXT: br label [[OMP_BODY_CONTINUE12:%.*]] 2730 // CHECK23: omp.body.continue12: 2731 // CHECK23-NEXT: br label [[OMP_INNER_FOR_INC13:%.*]] 2732 // CHECK23: omp.inner.for.inc13: 2733 // CHECK23-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2734 // CHECK23-NEXT: [[ADD14:%.*]] = add nsw i32 [[TMP18]], 1 2735 // CHECK23-NEXT: store i32 [[ADD14]], ptr [[DOTOMP_IV]], align 4 2736 // CHECK23-NEXT: br label [[OMP_INNER_FOR_COND3]], !llvm.loop [[LOOP10:![0-9]+]] 2737 // CHECK23: omp.inner.for.end15: 2738 // CHECK23-NEXT: br label [[OMP_IF_END]] 2739 // CHECK23: omp_if.end: 2740 // CHECK23-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 2741 // CHECK23: omp.loop.exit: 2742 // CHECK23-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 2743 // CHECK23-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 2744 // CHECK23-NEXT: [[TMP20:%.*]] = icmp ne i32 [[TMP19]], 0 2745 // CHECK23-NEXT: br i1 [[TMP20]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 2746 // CHECK23: .omp.final.then: 2747 // CHECK23-NEXT: store i32 123, ptr [[I]], align 4 2748 // CHECK23-NEXT: br label [[DOTOMP_FINAL_DONE]] 2749 // CHECK23: .omp.final.done: 2750 // CHECK23-NEXT: ret void 2751 // 2752 // 2753 // CHECK25-LABEL: define {{[^@]+}}@_Z21teams_template_structv 2754 // CHECK25-SAME: () #[[ATTR0:[0-9]+]] { 2755 // CHECK25-NEXT: entry: 2756 // CHECK25-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 2757 // CHECK25-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]]) 2758 // CHECK25-NEXT: ret i32 [[CALL]] 2759 // 2760 // 2761 // CHECK25-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 2762 // CHECK25-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat { 2763 // CHECK25-NEXT: entry: 2764 // CHECK25-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 2765 // CHECK25-NEXT: [[TMP:%.*]] = alloca i32, align 4 2766 // CHECK25-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2767 // CHECK25-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2768 // CHECK25-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2769 // CHECK25-NEXT: [[I:%.*]] = alloca i32, align 4 2770 // CHECK25-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 2771 // CHECK25-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 2772 // CHECK25-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2773 // CHECK25-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 2774 // CHECK25-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2775 // CHECK25-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 2776 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2777 // CHECK25: omp.inner.for.cond: 2778 // CHECK25-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2:![0-9]+]] 2779 // CHECK25-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP2]] 2780 // CHECK25-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] 2781 // CHECK25-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2782 // CHECK25: omp.inner.for.body: 2783 // CHECK25-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 2784 // CHECK25-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 2785 // CHECK25-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2786 // CHECK25-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP2]] 2787 // CHECK25-NEXT: [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 1 2788 // CHECK25-NEXT: [[TMP4:%.*]] = load float, ptr [[B]], align 4, !llvm.access.group [[ACC_GRP2]] 2789 // CHECK25-NEXT: [[CONV:%.*]] = fptosi float [[TMP4]] to i32 2790 // CHECK25-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 2791 // CHECK25-NEXT: [[TMP5:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP2]] 2792 // CHECK25-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP5]] to i64 2793 // CHECK25-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]] 2794 // CHECK25-NEXT: store i32 [[CONV]], ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP2]] 2795 // CHECK25-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2796 // CHECK25: omp.body.continue: 2797 // CHECK25-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2798 // CHECK25: omp.inner.for.inc: 2799 // CHECK25-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 2800 // CHECK25-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], 1 2801 // CHECK25-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 2802 // CHECK25-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] 2803 // CHECK25: omp.inner.for.end: 2804 // CHECK25-NEXT: store i32 123, ptr [[I]], align 4 2805 // CHECK25-NEXT: [[A3:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 2806 // CHECK25-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [123 x i32], ptr [[A3]], i64 0, i64 0 2807 // CHECK25-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4 2808 // CHECK25-NEXT: ret i32 [[TMP7]] 2809 // 2810 // 2811 // CHECK27-LABEL: define {{[^@]+}}@_Z21teams_template_structv 2812 // CHECK27-SAME: () #[[ATTR0:[0-9]+]] { 2813 // CHECK27-NEXT: entry: 2814 // CHECK27-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 2815 // CHECK27-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]]) 2816 // CHECK27-NEXT: ret i32 [[CALL]] 2817 // 2818 // 2819 // CHECK27-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 2820 // CHECK27-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 2821 // CHECK27-NEXT: entry: 2822 // CHECK27-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 2823 // CHECK27-NEXT: [[TMP:%.*]] = alloca i32, align 4 2824 // CHECK27-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2825 // CHECK27-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2826 // CHECK27-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2827 // CHECK27-NEXT: [[I:%.*]] = alloca i32, align 4 2828 // CHECK27-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 2829 // CHECK27-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 2830 // CHECK27-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2831 // CHECK27-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 2832 // CHECK27-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2833 // CHECK27-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 2834 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2835 // CHECK27: omp.inner.for.cond: 2836 // CHECK27-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3:![0-9]+]] 2837 // CHECK27-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP3]] 2838 // CHECK27-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] 2839 // CHECK27-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2840 // CHECK27: omp.inner.for.body: 2841 // CHECK27-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 2842 // CHECK27-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 2843 // CHECK27-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2844 // CHECK27-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]] 2845 // CHECK27-NEXT: [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 1 2846 // CHECK27-NEXT: [[TMP4:%.*]] = load float, ptr [[B]], align 4, !llvm.access.group [[ACC_GRP3]] 2847 // CHECK27-NEXT: [[CONV:%.*]] = fptosi float [[TMP4]] to i32 2848 // CHECK27-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 2849 // CHECK27-NEXT: [[TMP5:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]] 2850 // CHECK27-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP5]] 2851 // CHECK27-NEXT: store i32 [[CONV]], ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP3]] 2852 // CHECK27-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2853 // CHECK27: omp.body.continue: 2854 // CHECK27-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2855 // CHECK27: omp.inner.for.inc: 2856 // CHECK27-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 2857 // CHECK27-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP6]], 1 2858 // CHECK27-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 2859 // CHECK27-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] 2860 // CHECK27: omp.inner.for.end: 2861 // CHECK27-NEXT: store i32 123, ptr [[I]], align 4 2862 // CHECK27-NEXT: [[A3:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 2863 // CHECK27-NEXT: [[ARRAYIDX4:%.*]] = getelementptr inbounds [123 x i32], ptr [[A3]], i32 0, i32 0 2864 // CHECK27-NEXT: [[TMP7:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4 2865 // CHECK27-NEXT: ret i32 [[TMP7]] 2866 // 2867 // 2868 // CHECK29-LABEL: define {{[^@]+}}@_Z21teams_template_structv 2869 // CHECK29-SAME: () #[[ATTR0:[0-9]+]] { 2870 // CHECK29-NEXT: entry: 2871 // CHECK29-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 2872 // CHECK29-NEXT: [[CALL:%.*]] = call noundef signext i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]]) 2873 // CHECK29-NEXT: ret i32 [[CALL]] 2874 // 2875 // 2876 // CHECK29-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 2877 // CHECK29-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat { 2878 // CHECK29-NEXT: entry: 2879 // CHECK29-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 8 2880 // CHECK29-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 2881 // CHECK29-NEXT: [[TMP:%.*]] = alloca i32, align 4 2882 // CHECK29-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2883 // CHECK29-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2884 // CHECK29-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2885 // CHECK29-NEXT: [[I:%.*]] = alloca i32, align 4 2886 // CHECK29-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 8 2887 // CHECK29-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8 2888 // CHECK29-NEXT: [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 1 2889 // CHECK29-NEXT: [[TMP0:%.*]] = load float, ptr [[B]], align 4 2890 // CHECK29-NEXT: [[TOBOOL:%.*]] = fcmp une float [[TMP0]], 0.000000e+00 2891 // CHECK29-NEXT: [[STOREDV:%.*]] = zext i1 [[TOBOOL]] to i8 2892 // CHECK29-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR_]], align 1 2893 // CHECK29-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2894 // CHECK29-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 2895 // CHECK29-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2896 // CHECK29-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_IV]], align 4 2897 // CHECK29-NEXT: [[TMP2:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1 2898 // CHECK29-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP2]] to i1 2899 // CHECK29-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] 2900 // CHECK29: omp_if.then: 2901 // CHECK29-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 2902 // CHECK29: omp.inner.for.cond: 2903 // CHECK29-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2:![0-9]+]] 2904 // CHECK29-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP2]] 2905 // CHECK29-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP3]], [[TMP4]] 2906 // CHECK29-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 2907 // CHECK29: omp.inner.for.body: 2908 // CHECK29-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 2909 // CHECK29-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP5]], 1 2910 // CHECK29-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 2911 // CHECK29-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP2]] 2912 // CHECK29-NEXT: [[B2:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 1 2913 // CHECK29-NEXT: [[TMP6:%.*]] = load float, ptr [[B2]], align 4, !nontemporal [[META3:![0-9]+]], !llvm.access.group [[ACC_GRP2]] 2914 // CHECK29-NEXT: [[CONV:%.*]] = fptosi float [[TMP6]] to i32 2915 // CHECK29-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 2916 // CHECK29-NEXT: [[TMP7:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP2]] 2917 // CHECK29-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP7]] to i64 2918 // CHECK29-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]] 2919 // CHECK29-NEXT: store i32 [[CONV]], ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP2]] 2920 // CHECK29-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 2921 // CHECK29: omp.body.continue: 2922 // CHECK29-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 2923 // CHECK29: omp.inner.for.inc: 2924 // CHECK29-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 2925 // CHECK29-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP8]], 1 2926 // CHECK29-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 2927 // CHECK29-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] 2928 // CHECK29: omp.inner.for.end: 2929 // CHECK29-NEXT: br label [[OMP_IF_END:%.*]] 2930 // CHECK29: omp_if.else: 2931 // CHECK29-NEXT: br label [[OMP_INNER_FOR_COND4:%.*]] 2932 // CHECK29: omp.inner.for.cond4: 2933 // CHECK29-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2934 // CHECK29-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 2935 // CHECK29-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]] 2936 // CHECK29-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY6:%.*]], label [[OMP_INNER_FOR_END17:%.*]] 2937 // CHECK29: omp.inner.for.body6: 2938 // CHECK29-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2939 // CHECK29-NEXT: [[MUL7:%.*]] = mul nsw i32 [[TMP11]], 1 2940 // CHECK29-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL7]] 2941 // CHECK29-NEXT: store i32 [[ADD8]], ptr [[I]], align 4 2942 // CHECK29-NEXT: [[B9:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 1 2943 // CHECK29-NEXT: [[TMP12:%.*]] = load float, ptr [[B9]], align 4 2944 // CHECK29-NEXT: [[CONV10:%.*]] = fptosi float [[TMP12]] to i32 2945 // CHECK29-NEXT: [[A11:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 2946 // CHECK29-NEXT: [[TMP13:%.*]] = load i32, ptr [[I]], align 4 2947 // CHECK29-NEXT: [[IDXPROM12:%.*]] = sext i32 [[TMP13]] to i64 2948 // CHECK29-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds [123 x i32], ptr [[A11]], i64 0, i64 [[IDXPROM12]] 2949 // CHECK29-NEXT: store i32 [[CONV10]], ptr [[ARRAYIDX13]], align 4 2950 // CHECK29-NEXT: br label [[OMP_BODY_CONTINUE14:%.*]] 2951 // CHECK29: omp.body.continue14: 2952 // CHECK29-NEXT: br label [[OMP_INNER_FOR_INC15:%.*]] 2953 // CHECK29: omp.inner.for.inc15: 2954 // CHECK29-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 2955 // CHECK29-NEXT: [[ADD16:%.*]] = add nsw i32 [[TMP14]], 1 2956 // CHECK29-NEXT: store i32 [[ADD16]], ptr [[DOTOMP_IV]], align 4 2957 // CHECK29-NEXT: br label [[OMP_INNER_FOR_COND4]], !llvm.loop [[LOOP7:![0-9]+]] 2958 // CHECK29: omp.inner.for.end17: 2959 // CHECK29-NEXT: br label [[OMP_IF_END]] 2960 // CHECK29: omp_if.end: 2961 // CHECK29-NEXT: store i32 123, ptr [[I]], align 4 2962 // CHECK29-NEXT: [[A18:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 2963 // CHECK29-NEXT: [[ARRAYIDX19:%.*]] = getelementptr inbounds [123 x i32], ptr [[A18]], i64 0, i64 0 2964 // CHECK29-NEXT: [[TMP15:%.*]] = load i32, ptr [[ARRAYIDX19]], align 4 2965 // CHECK29-NEXT: ret i32 [[TMP15]] 2966 // 2967 // 2968 // CHECK31-LABEL: define {{[^@]+}}@_Z21teams_template_structv 2969 // CHECK31-SAME: () #[[ATTR0:[0-9]+]] { 2970 // CHECK31-NEXT: entry: 2971 // CHECK31-NEXT: [[V:%.*]] = alloca [[STRUCT_SS:%.*]], align 4 2972 // CHECK31-NEXT: [[CALL:%.*]] = call noundef i32 @_ZN2SSIiLi123ELx456EE3fooEv(ptr noundef nonnull align 4 dereferenceable(496) [[V]]) 2973 // CHECK31-NEXT: ret i32 [[CALL]] 2974 // 2975 // 2976 // CHECK31-LABEL: define {{[^@]+}}@_ZN2SSIiLi123ELx456EE3fooEv 2977 // CHECK31-SAME: (ptr noundef nonnull align 4 dereferenceable(496) [[THIS:%.*]]) #[[ATTR0]] comdat align 2 { 2978 // CHECK31-NEXT: entry: 2979 // CHECK31-NEXT: [[THIS_ADDR:%.*]] = alloca ptr, align 4 2980 // CHECK31-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 2981 // CHECK31-NEXT: [[TMP:%.*]] = alloca i32, align 4 2982 // CHECK31-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 2983 // CHECK31-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 2984 // CHECK31-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 2985 // CHECK31-NEXT: [[I:%.*]] = alloca i32, align 4 2986 // CHECK31-NEXT: store ptr [[THIS]], ptr [[THIS_ADDR]], align 4 2987 // CHECK31-NEXT: [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 4 2988 // CHECK31-NEXT: [[B:%.*]] = getelementptr inbounds nuw [[STRUCT_SS:%.*]], ptr [[THIS1]], i32 0, i32 1 2989 // CHECK31-NEXT: [[TMP0:%.*]] = load float, ptr [[B]], align 4 2990 // CHECK31-NEXT: [[TOBOOL:%.*]] = fcmp une float [[TMP0]], 0.000000e+00 2991 // CHECK31-NEXT: [[STOREDV:%.*]] = zext i1 [[TOBOOL]] to i8 2992 // CHECK31-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR_]], align 1 2993 // CHECK31-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 2994 // CHECK31-NEXT: store i32 122, ptr [[DOTOMP_UB]], align 4 2995 // CHECK31-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 2996 // CHECK31-NEXT: store i32 [[TMP1]], ptr [[DOTOMP_IV]], align 4 2997 // CHECK31-NEXT: [[TMP2:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1 2998 // CHECK31-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP2]] to i1 2999 // CHECK31-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] 3000 // CHECK31: omp_if.then: 3001 // CHECK31-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3002 // CHECK31: omp.inner.for.cond: 3003 // CHECK31-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3:![0-9]+]] 3004 // CHECK31-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP3]] 3005 // CHECK31-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP3]], [[TMP4]] 3006 // CHECK31-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3007 // CHECK31: omp.inner.for.body: 3008 // CHECK31-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 3009 // CHECK31-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP5]], 1 3010 // CHECK31-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3011 // CHECK31-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]] 3012 // CHECK31-NEXT: [[B2:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 1 3013 // CHECK31-NEXT: [[TMP6:%.*]] = load float, ptr [[B2]], align 4, !nontemporal [[META4:![0-9]+]], !llvm.access.group [[ACC_GRP3]] 3014 // CHECK31-NEXT: [[CONV:%.*]] = fptosi float [[TMP6]] to i32 3015 // CHECK31-NEXT: [[A:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 3016 // CHECK31-NEXT: [[TMP7:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP3]] 3017 // CHECK31-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [123 x i32], ptr [[A]], i32 0, i32 [[TMP7]] 3018 // CHECK31-NEXT: store i32 [[CONV]], ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP3]] 3019 // CHECK31-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3020 // CHECK31: omp.body.continue: 3021 // CHECK31-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3022 // CHECK31: omp.inner.for.inc: 3023 // CHECK31-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 3024 // CHECK31-NEXT: [[ADD3:%.*]] = add nsw i32 [[TMP8]], 1 3025 // CHECK31-NEXT: store i32 [[ADD3]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 3026 // CHECK31-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP5:![0-9]+]] 3027 // CHECK31: omp.inner.for.end: 3028 // CHECK31-NEXT: br label [[OMP_IF_END:%.*]] 3029 // CHECK31: omp_if.else: 3030 // CHECK31-NEXT: br label [[OMP_INNER_FOR_COND4:%.*]] 3031 // CHECK31: omp.inner.for.cond4: 3032 // CHECK31-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3033 // CHECK31-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3034 // CHECK31-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]] 3035 // CHECK31-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY6:%.*]], label [[OMP_INNER_FOR_END16:%.*]] 3036 // CHECK31: omp.inner.for.body6: 3037 // CHECK31-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3038 // CHECK31-NEXT: [[MUL7:%.*]] = mul nsw i32 [[TMP11]], 1 3039 // CHECK31-NEXT: [[ADD8:%.*]] = add nsw i32 0, [[MUL7]] 3040 // CHECK31-NEXT: store i32 [[ADD8]], ptr [[I]], align 4 3041 // CHECK31-NEXT: [[B9:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 1 3042 // CHECK31-NEXT: [[TMP12:%.*]] = load float, ptr [[B9]], align 4 3043 // CHECK31-NEXT: [[CONV10:%.*]] = fptosi float [[TMP12]] to i32 3044 // CHECK31-NEXT: [[A11:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 3045 // CHECK31-NEXT: [[TMP13:%.*]] = load i32, ptr [[I]], align 4 3046 // CHECK31-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds [123 x i32], ptr [[A11]], i32 0, i32 [[TMP13]] 3047 // CHECK31-NEXT: store i32 [[CONV10]], ptr [[ARRAYIDX12]], align 4 3048 // CHECK31-NEXT: br label [[OMP_BODY_CONTINUE13:%.*]] 3049 // CHECK31: omp.body.continue13: 3050 // CHECK31-NEXT: br label [[OMP_INNER_FOR_INC14:%.*]] 3051 // CHECK31: omp.inner.for.inc14: 3052 // CHECK31-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 3053 // CHECK31-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP14]], 1 3054 // CHECK31-NEXT: store i32 [[ADD15]], ptr [[DOTOMP_IV]], align 4 3055 // CHECK31-NEXT: br label [[OMP_INNER_FOR_COND4]], !llvm.loop [[LOOP8:![0-9]+]] 3056 // CHECK31: omp.inner.for.end16: 3057 // CHECK31-NEXT: br label [[OMP_IF_END]] 3058 // CHECK31: omp_if.end: 3059 // CHECK31-NEXT: store i32 123, ptr [[I]], align 4 3060 // CHECK31-NEXT: [[A17:%.*]] = getelementptr inbounds nuw [[STRUCT_SS]], ptr [[THIS1]], i32 0, i32 0 3061 // CHECK31-NEXT: [[ARRAYIDX18:%.*]] = getelementptr inbounds [123 x i32], ptr [[A17]], i32 0, i32 0 3062 // CHECK31-NEXT: [[TMP15:%.*]] = load i32, ptr [[ARRAYIDX18]], align 4 3063 // CHECK31-NEXT: ret i32 [[TMP15]] 3064 // 3065 // 3066 // CHECK33-LABEL: define {{[^@]+}}@main 3067 // CHECK33-SAME: (i32 noundef signext [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 3068 // CHECK33-NEXT: entry: 3069 // CHECK33-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 3070 // CHECK33-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 3071 // CHECK33-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 8 3072 // CHECK33-NEXT: [[N:%.*]] = alloca i32, align 4 3073 // CHECK33-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 8 3074 // CHECK33-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 3075 // CHECK33-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 3076 // CHECK33-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8 3077 // CHECK33-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8 3078 // CHECK33-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8 3079 // CHECK33-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 8 3080 // CHECK33-NEXT: [[TMP:%.*]] = alloca i32, align 4 3081 // CHECK33-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3082 // CHECK33-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3083 // CHECK33-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 3084 // CHECK33-NEXT: store i32 0, ptr [[RETVAL]], align 4 3085 // CHECK33-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 3086 // CHECK33-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8 3087 // CHECK33-NEXT: store i32 100, ptr [[N]], align 4 3088 // CHECK33-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 3089 // CHECK33-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 3090 // CHECK33-NEXT: [[TMP2:%.*]] = call ptr @llvm.stacksave.p0() 3091 // CHECK33-NEXT: store ptr [[TMP2]], ptr [[SAVED_STACK]], align 8 3092 // CHECK33-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 3093 // CHECK33-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8 3094 // CHECK33-NEXT: [[TMP3:%.*]] = load i32, ptr [[N]], align 4 3095 // CHECK33-NEXT: store i32 [[TMP3]], ptr [[N_CASTED]], align 4 3096 // CHECK33-NEXT: [[TMP4:%.*]] = load i64, ptr [[N_CASTED]], align 8 3097 // CHECK33-NEXT: [[TMP5:%.*]] = mul nuw i64 [[TMP1]], 4 3098 // CHECK33-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 24, i1 false) 3099 // CHECK33-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3100 // CHECK33-NEXT: store i64 [[TMP4]], ptr [[TMP6]], align 8 3101 // CHECK33-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3102 // CHECK33-NEXT: store i64 [[TMP4]], ptr [[TMP7]], align 8 3103 // CHECK33-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 3104 // CHECK33-NEXT: store ptr null, ptr [[TMP8]], align 8 3105 // CHECK33-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 3106 // CHECK33-NEXT: store i64 [[TMP1]], ptr [[TMP9]], align 8 3107 // CHECK33-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 3108 // CHECK33-NEXT: store i64 [[TMP1]], ptr [[TMP10]], align 8 3109 // CHECK33-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 3110 // CHECK33-NEXT: store ptr null, ptr [[TMP11]], align 8 3111 // CHECK33-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 3112 // CHECK33-NEXT: store ptr [[VLA]], ptr [[TMP12]], align 8 3113 // CHECK33-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 3114 // CHECK33-NEXT: store ptr [[VLA]], ptr [[TMP13]], align 8 3115 // CHECK33-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2 3116 // CHECK33-NEXT: store i64 [[TMP5]], ptr [[TMP14]], align 8 3117 // CHECK33-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 3118 // CHECK33-NEXT: store ptr null, ptr [[TMP15]], align 8 3119 // CHECK33-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3120 // CHECK33-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3121 // CHECK33-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 3122 // CHECK33-NEXT: [[TMP19:%.*]] = load i32, ptr [[N]], align 4 3123 // CHECK33-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4 3124 // CHECK33-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3125 // CHECK33-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0 3126 // CHECK33-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3127 // CHECK33-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3128 // CHECK33-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 3129 // CHECK33-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3130 // CHECK33-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], 1 3131 // CHECK33-NEXT: [[TMP22:%.*]] = zext i32 [[ADD]] to i64 3132 // CHECK33-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 3133 // CHECK33-NEXT: store i32 3, ptr [[TMP23]], align 4 3134 // CHECK33-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 3135 // CHECK33-NEXT: store i32 3, ptr [[TMP24]], align 4 3136 // CHECK33-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 3137 // CHECK33-NEXT: store ptr [[TMP16]], ptr [[TMP25]], align 8 3138 // CHECK33-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 3139 // CHECK33-NEXT: store ptr [[TMP17]], ptr [[TMP26]], align 8 3140 // CHECK33-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 3141 // CHECK33-NEXT: store ptr [[TMP18]], ptr [[TMP27]], align 8 3142 // CHECK33-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 3143 // CHECK33-NEXT: store ptr @.offload_maptypes, ptr [[TMP28]], align 8 3144 // CHECK33-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 3145 // CHECK33-NEXT: store ptr null, ptr [[TMP29]], align 8 3146 // CHECK33-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 3147 // CHECK33-NEXT: store ptr null, ptr [[TMP30]], align 8 3148 // CHECK33-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 3149 // CHECK33-NEXT: store i64 [[TMP22]], ptr [[TMP31]], align 8 3150 // CHECK33-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 3151 // CHECK33-NEXT: store i64 0, ptr [[TMP32]], align 8 3152 // CHECK33-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 3153 // CHECK33-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4 3154 // CHECK33-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 3155 // CHECK33-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP34]], align 4 3156 // CHECK33-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 3157 // CHECK33-NEXT: store i32 0, ptr [[TMP35]], align 4 3158 // CHECK33-NEXT: [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, ptr [[KERNEL_ARGS]]) 3159 // CHECK33-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0 3160 // CHECK33-NEXT: br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 3161 // CHECK33: omp_offload.failed: 3162 // CHECK33-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i64 [[TMP4]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3:[0-9]+]] 3163 // CHECK33-NEXT: br label [[OMP_OFFLOAD_CONT]] 3164 // CHECK33: omp_offload.cont: 3165 // CHECK33-NEXT: [[TMP38:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 3166 // CHECK33-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10EEiT_(i32 noundef signext [[TMP38]]) 3167 // CHECK33-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 3168 // CHECK33-NEXT: [[TMP39:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 3169 // CHECK33-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP39]]) 3170 // CHECK33-NEXT: [[TMP40:%.*]] = load i32, ptr [[RETVAL]], align 4 3171 // CHECK33-NEXT: ret i32 [[TMP40]] 3172 // 3173 // 3174 // CHECK33-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192 3175 // CHECK33-SAME: (i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 3176 // CHECK33-NEXT: entry: 3177 // CHECK33-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 3178 // CHECK33-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 3179 // CHECK33-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 3180 // CHECK33-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 3181 // CHECK33-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 3182 // CHECK33-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 3183 // CHECK33-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 3184 // CHECK33-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 3185 // CHECK33-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]]) 3186 // CHECK33-NEXT: ret void 3187 // 3188 // 3189 // CHECK33-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.omp_outlined 3190 // CHECK33-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 3191 // CHECK33-NEXT: entry: 3192 // CHECK33-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 3193 // CHECK33-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 3194 // CHECK33-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8 3195 // CHECK33-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 3196 // CHECK33-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 3197 // CHECK33-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3198 // CHECK33-NEXT: [[TMP:%.*]] = alloca i32, align 4 3199 // CHECK33-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3200 // CHECK33-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3201 // CHECK33-NEXT: [[I:%.*]] = alloca i32, align 4 3202 // CHECK33-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3203 // CHECK33-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3204 // CHECK33-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3205 // CHECK33-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3206 // CHECK33-NEXT: [[I3:%.*]] = alloca i32, align 4 3207 // CHECK33-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 3208 // CHECK33-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 3209 // CHECK33-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8 3210 // CHECK33-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 3211 // CHECK33-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 3212 // CHECK33-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8 3213 // CHECK33-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 3214 // CHECK33-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 3215 // CHECK33-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 3216 // CHECK33-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 3217 // CHECK33-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3218 // CHECK33-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 3219 // CHECK33-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3220 // CHECK33-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3221 // CHECK33-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 3222 // CHECK33-NEXT: store i32 0, ptr [[I]], align 4 3223 // CHECK33-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3224 // CHECK33-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 3225 // CHECK33-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3226 // CHECK33: omp.precond.then: 3227 // CHECK33-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 3228 // CHECK33-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3229 // CHECK33-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 3230 // CHECK33-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 3231 // CHECK33-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 3232 // CHECK33-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 3233 // CHECK33-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 3234 // CHECK33-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 3235 // CHECK33-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3236 // CHECK33-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3237 // CHECK33-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 3238 // CHECK33-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3239 // CHECK33: cond.true: 3240 // CHECK33-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3241 // CHECK33-NEXT: br label [[COND_END:%.*]] 3242 // CHECK33: cond.false: 3243 // CHECK33-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3244 // CHECK33-NEXT: br label [[COND_END]] 3245 // CHECK33: cond.end: 3246 // CHECK33-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 3247 // CHECK33-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 3248 // CHECK33-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 3249 // CHECK33-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 3250 // CHECK33-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3251 // CHECK33: omp.inner.for.cond: 3252 // CHECK33-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5:![0-9]+]] 3253 // CHECK33-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP5]] 3254 // CHECK33-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 3255 // CHECK33-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3256 // CHECK33: omp.inner.for.body: 3257 // CHECK33-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 3258 // CHECK33-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 3259 // CHECK33-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3260 // CHECK33-NEXT: store i32 [[ADD]], ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP5]] 3261 // CHECK33-NEXT: [[TMP17:%.*]] = load i32, ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP5]] 3262 // CHECK33-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP17]] to i64 3263 // CHECK33-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]] 3264 // CHECK33-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP5]] 3265 // CHECK33-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3266 // CHECK33: omp.body.continue: 3267 // CHECK33-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3268 // CHECK33: omp.inner.for.inc: 3269 // CHECK33-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 3270 // CHECK33-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], 1 3271 // CHECK33-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 3272 // CHECK33-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] 3273 // CHECK33: omp.inner.for.end: 3274 // CHECK33-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3275 // CHECK33: omp.loop.exit: 3276 // CHECK33-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 3277 // CHECK33-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4 3278 // CHECK33-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP20]]) 3279 // CHECK33-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 3280 // CHECK33-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 3281 // CHECK33-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 3282 // CHECK33: .omp.final.then: 3283 // CHECK33-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3284 // CHECK33-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP23]], 0 3285 // CHECK33-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 3286 // CHECK33-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1 3287 // CHECK33-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]] 3288 // CHECK33-NEXT: store i32 [[ADD10]], ptr [[I3]], align 4 3289 // CHECK33-NEXT: br label [[DOTOMP_FINAL_DONE]] 3290 // CHECK33: .omp.final.done: 3291 // CHECK33-NEXT: br label [[OMP_PRECOND_END]] 3292 // CHECK33: omp.precond.end: 3293 // CHECK33-NEXT: ret void 3294 // 3295 // 3296 // CHECK33-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 3297 // CHECK33-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 3298 // CHECK33-NEXT: entry: 3299 // CHECK33-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 3300 // CHECK33-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 3301 // CHECK33-NEXT: [[TE:%.*]] = alloca i32, align 4 3302 // CHECK33-NEXT: [[TH:%.*]] = alloca i32, align 4 3303 // CHECK33-NEXT: [[TE_CASTED:%.*]] = alloca i64, align 8 3304 // CHECK33-NEXT: [[TH_CASTED:%.*]] = alloca i64, align 8 3305 // CHECK33-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8 3306 // CHECK33-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8 3307 // CHECK33-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8 3308 // CHECK33-NEXT: [[TMP:%.*]] = alloca i32, align 4 3309 // CHECK33-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 3310 // CHECK33-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 3311 // CHECK33-NEXT: store i32 0, ptr [[TE]], align 4 3312 // CHECK33-NEXT: store i32 128, ptr [[TH]], align 4 3313 // CHECK33-NEXT: [[TMP0:%.*]] = load i32, ptr [[TE]], align 4 3314 // CHECK33-NEXT: store i32 [[TMP0]], ptr [[TE_CASTED]], align 4 3315 // CHECK33-NEXT: [[TMP1:%.*]] = load i64, ptr [[TE_CASTED]], align 8 3316 // CHECK33-NEXT: [[TMP2:%.*]] = load i32, ptr [[TH]], align 4 3317 // CHECK33-NEXT: store i32 [[TMP2]], ptr [[TH_CASTED]], align 4 3318 // CHECK33-NEXT: [[TMP3:%.*]] = load i64, ptr [[TH_CASTED]], align 8 3319 // CHECK33-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3320 // CHECK33-NEXT: store i64 [[TMP1]], ptr [[TMP4]], align 8 3321 // CHECK33-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3322 // CHECK33-NEXT: store i64 [[TMP1]], ptr [[TMP5]], align 8 3323 // CHECK33-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 3324 // CHECK33-NEXT: store ptr null, ptr [[TMP6]], align 8 3325 // CHECK33-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 3326 // CHECK33-NEXT: store i64 [[TMP3]], ptr [[TMP7]], align 8 3327 // CHECK33-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 3328 // CHECK33-NEXT: store i64 [[TMP3]], ptr [[TMP8]], align 8 3329 // CHECK33-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 3330 // CHECK33-NEXT: store ptr null, ptr [[TMP9]], align 8 3331 // CHECK33-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 3332 // CHECK33-NEXT: store ptr [[A]], ptr [[TMP10]], align 8 3333 // CHECK33-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 3334 // CHECK33-NEXT: store ptr [[A]], ptr [[TMP11]], align 8 3335 // CHECK33-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 3336 // CHECK33-NEXT: store ptr null, ptr [[TMP12]], align 8 3337 // CHECK33-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3338 // CHECK33-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3339 // CHECK33-NEXT: [[TMP15:%.*]] = load i32, ptr [[TE]], align 4 3340 // CHECK33-NEXT: [[TMP16:%.*]] = load i32, ptr [[TH]], align 4 3341 // CHECK33-NEXT: [[TMP17:%.*]] = icmp ult i32 [[TMP16]], 1 3342 // CHECK33-NEXT: [[TMP18:%.*]] = select i1 [[TMP17]], i32 [[TMP16]], i32 1 3343 // CHECK33-NEXT: [[TMP19:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP15]], 0 3344 // CHECK33-NEXT: [[TMP20:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP18]], 0 3345 // CHECK33-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 3346 // CHECK33-NEXT: store i32 3, ptr [[TMP21]], align 4 3347 // CHECK33-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 3348 // CHECK33-NEXT: store i32 3, ptr [[TMP22]], align 4 3349 // CHECK33-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 3350 // CHECK33-NEXT: store ptr [[TMP13]], ptr [[TMP23]], align 8 3351 // CHECK33-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 3352 // CHECK33-NEXT: store ptr [[TMP14]], ptr [[TMP24]], align 8 3353 // CHECK33-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 3354 // CHECK33-NEXT: store ptr @.offload_sizes.1, ptr [[TMP25]], align 8 3355 // CHECK33-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 3356 // CHECK33-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP26]], align 8 3357 // CHECK33-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 3358 // CHECK33-NEXT: store ptr null, ptr [[TMP27]], align 8 3359 // CHECK33-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 3360 // CHECK33-NEXT: store ptr null, ptr [[TMP28]], align 8 3361 // CHECK33-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 3362 // CHECK33-NEXT: store i64 10, ptr [[TMP29]], align 8 3363 // CHECK33-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 3364 // CHECK33-NEXT: store i64 0, ptr [[TMP30]], align 8 3365 // CHECK33-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 3366 // CHECK33-NEXT: store [3 x i32] [[TMP19]], ptr [[TMP31]], align 4 3367 // CHECK33-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 3368 // CHECK33-NEXT: store [3 x i32] [[TMP20]], ptr [[TMP32]], align 4 3369 // CHECK33-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 3370 // CHECK33-NEXT: store i32 0, ptr [[TMP33]], align 4 3371 // CHECK33-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 [[TMP15]], i32 [[TMP18]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, ptr [[KERNEL_ARGS]]) 3372 // CHECK33-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0 3373 // CHECK33-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 3374 // CHECK33: omp_offload.failed: 3375 // CHECK33-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i64 [[TMP1]], i64 [[TMP3]], ptr [[A]]) #[[ATTR3]] 3376 // CHECK33-NEXT: br label [[OMP_OFFLOAD_CONT]] 3377 // CHECK33: omp_offload.cont: 3378 // CHECK33-NEXT: ret i32 0 3379 // 3380 // 3381 // CHECK33-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181 3382 // CHECK33-SAME: (i64 noundef [[TE:%.*]], i64 noundef [[TH:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 3383 // CHECK33-NEXT: entry: 3384 // CHECK33-NEXT: [[TE_ADDR:%.*]] = alloca i64, align 8 3385 // CHECK33-NEXT: [[TH_ADDR:%.*]] = alloca i64, align 8 3386 // CHECK33-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 3387 // CHECK33-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2]]) 3388 // CHECK33-NEXT: store i64 [[TE]], ptr [[TE_ADDR]], align 8 3389 // CHECK33-NEXT: store i64 [[TH]], ptr [[TH_ADDR]], align 8 3390 // CHECK33-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 3391 // CHECK33-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 3392 // CHECK33-NEXT: [[TMP2:%.*]] = load i32, ptr [[TE_ADDR]], align 4 3393 // CHECK33-NEXT: [[TMP3:%.*]] = load i32, ptr [[TH_ADDR]], align 4 3394 // CHECK33-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB2]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 3395 // CHECK33-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.omp_outlined, ptr [[TMP1]]) 3396 // CHECK33-NEXT: ret void 3397 // 3398 // 3399 // CHECK33-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.omp_outlined 3400 // CHECK33-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 3401 // CHECK33-NEXT: entry: 3402 // CHECK33-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 3403 // CHECK33-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 3404 // CHECK33-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 3405 // CHECK33-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3406 // CHECK33-NEXT: [[TMP:%.*]] = alloca i32, align 4 3407 // CHECK33-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3408 // CHECK33-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3409 // CHECK33-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3410 // CHECK33-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3411 // CHECK33-NEXT: [[I:%.*]] = alloca i32, align 4 3412 // CHECK33-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 3413 // CHECK33-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 3414 // CHECK33-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 3415 // CHECK33-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 3416 // CHECK33-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 3417 // CHECK33-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 3418 // CHECK33-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 3419 // CHECK33-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 3420 // CHECK33-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 3421 // CHECK33-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 3422 // CHECK33-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 3423 // CHECK33-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3424 // CHECK33-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 3425 // CHECK33-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3426 // CHECK33: cond.true: 3427 // CHECK33-NEXT: br label [[COND_END:%.*]] 3428 // CHECK33: cond.false: 3429 // CHECK33-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3430 // CHECK33-NEXT: br label [[COND_END]] 3431 // CHECK33: cond.end: 3432 // CHECK33-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 3433 // CHECK33-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 3434 // CHECK33-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 3435 // CHECK33-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 3436 // CHECK33-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3437 // CHECK33: omp.inner.for.cond: 3438 // CHECK33-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11:![0-9]+]] 3439 // CHECK33-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP11]] 3440 // CHECK33-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 3441 // CHECK33-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3442 // CHECK33: omp.inner.for.body: 3443 // CHECK33-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]] 3444 // CHECK33-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 3445 // CHECK33-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3446 // CHECK33-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP11]] 3447 // CHECK33-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP11]] 3448 // CHECK33-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP9]] to i64 3449 // CHECK33-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]] 3450 // CHECK33-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP11]] 3451 // CHECK33-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3452 // CHECK33: omp.body.continue: 3453 // CHECK33-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3454 // CHECK33: omp.inner.for.inc: 3455 // CHECK33-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]] 3456 // CHECK33-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 3457 // CHECK33-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP11]] 3458 // CHECK33-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP12:![0-9]+]] 3459 // CHECK33: omp.inner.for.end: 3460 // CHECK33-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3461 // CHECK33: omp.loop.exit: 3462 // CHECK33-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 3463 // CHECK33-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 3464 // CHECK33-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0 3465 // CHECK33-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 3466 // CHECK33: .omp.final.then: 3467 // CHECK33-NEXT: store i32 10, ptr [[I]], align 4 3468 // CHECK33-NEXT: br label [[DOTOMP_FINAL_DONE]] 3469 // CHECK33: .omp.final.done: 3470 // CHECK33-NEXT: ret void 3471 // 3472 // 3473 // CHECK35-LABEL: define {{[^@]+}}@main 3474 // CHECK35-SAME: (i32 noundef [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 3475 // CHECK35-NEXT: entry: 3476 // CHECK35-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 3477 // CHECK35-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 3478 // CHECK35-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 4 3479 // CHECK35-NEXT: [[N:%.*]] = alloca i32, align 4 3480 // CHECK35-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 4 3481 // CHECK35-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 3482 // CHECK35-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 3483 // CHECK35-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4 3484 // CHECK35-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4 3485 // CHECK35-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4 3486 // CHECK35-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [3 x i64], align 4 3487 // CHECK35-NEXT: [[TMP:%.*]] = alloca i32, align 4 3488 // CHECK35-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3489 // CHECK35-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3490 // CHECK35-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 3491 // CHECK35-NEXT: store i32 0, ptr [[RETVAL]], align 4 3492 // CHECK35-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 3493 // CHECK35-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 4 3494 // CHECK35-NEXT: store i32 100, ptr [[N]], align 4 3495 // CHECK35-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 3496 // CHECK35-NEXT: [[TMP1:%.*]] = call ptr @llvm.stacksave.p0() 3497 // CHECK35-NEXT: store ptr [[TMP1]], ptr [[SAVED_STACK]], align 4 3498 // CHECK35-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 3499 // CHECK35-NEXT: store i32 [[TMP0]], ptr [[__VLA_EXPR0]], align 4 3500 // CHECK35-NEXT: [[TMP2:%.*]] = load i32, ptr [[N]], align 4 3501 // CHECK35-NEXT: store i32 [[TMP2]], ptr [[N_CASTED]], align 4 3502 // CHECK35-NEXT: [[TMP3:%.*]] = load i32, ptr [[N_CASTED]], align 4 3503 // CHECK35-NEXT: [[TMP4:%.*]] = mul nuw i32 [[TMP0]], 4 3504 // CHECK35-NEXT: [[TMP5:%.*]] = sext i32 [[TMP4]] to i64 3505 // CHECK35-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 24, i1 false) 3506 // CHECK35-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3507 // CHECK35-NEXT: store i32 [[TMP3]], ptr [[TMP6]], align 4 3508 // CHECK35-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3509 // CHECK35-NEXT: store i32 [[TMP3]], ptr [[TMP7]], align 4 3510 // CHECK35-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 3511 // CHECK35-NEXT: store ptr null, ptr [[TMP8]], align 4 3512 // CHECK35-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 3513 // CHECK35-NEXT: store i32 [[TMP0]], ptr [[TMP9]], align 4 3514 // CHECK35-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 3515 // CHECK35-NEXT: store i32 [[TMP0]], ptr [[TMP10]], align 4 3516 // CHECK35-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 3517 // CHECK35-NEXT: store ptr null, ptr [[TMP11]], align 4 3518 // CHECK35-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 3519 // CHECK35-NEXT: store ptr [[VLA]], ptr [[TMP12]], align 4 3520 // CHECK35-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 3521 // CHECK35-NEXT: store ptr [[VLA]], ptr [[TMP13]], align 4 3522 // CHECK35-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 2 3523 // CHECK35-NEXT: store i64 [[TMP5]], ptr [[TMP14]], align 4 3524 // CHECK35-NEXT: [[TMP15:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 3525 // CHECK35-NEXT: store ptr null, ptr [[TMP15]], align 4 3526 // CHECK35-NEXT: [[TMP16:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3527 // CHECK35-NEXT: [[TMP17:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3528 // CHECK35-NEXT: [[TMP18:%.*]] = getelementptr inbounds [3 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 3529 // CHECK35-NEXT: [[TMP19:%.*]] = load i32, ptr [[N]], align 4 3530 // CHECK35-NEXT: store i32 [[TMP19]], ptr [[DOTCAPTURE_EXPR_]], align 4 3531 // CHECK35-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3532 // CHECK35-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP20]], 0 3533 // CHECK35-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3534 // CHECK35-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3535 // CHECK35-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 3536 // CHECK35-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3537 // CHECK35-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP21]], 1 3538 // CHECK35-NEXT: [[TMP22:%.*]] = zext i32 [[ADD]] to i64 3539 // CHECK35-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 3540 // CHECK35-NEXT: store i32 3, ptr [[TMP23]], align 4 3541 // CHECK35-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 3542 // CHECK35-NEXT: store i32 3, ptr [[TMP24]], align 4 3543 // CHECK35-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 3544 // CHECK35-NEXT: store ptr [[TMP16]], ptr [[TMP25]], align 4 3545 // CHECK35-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 3546 // CHECK35-NEXT: store ptr [[TMP17]], ptr [[TMP26]], align 4 3547 // CHECK35-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 3548 // CHECK35-NEXT: store ptr [[TMP18]], ptr [[TMP27]], align 4 3549 // CHECK35-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 3550 // CHECK35-NEXT: store ptr @.offload_maptypes, ptr [[TMP28]], align 4 3551 // CHECK35-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 3552 // CHECK35-NEXT: store ptr null, ptr [[TMP29]], align 4 3553 // CHECK35-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 3554 // CHECK35-NEXT: store ptr null, ptr [[TMP30]], align 4 3555 // CHECK35-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 3556 // CHECK35-NEXT: store i64 [[TMP22]], ptr [[TMP31]], align 8 3557 // CHECK35-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 3558 // CHECK35-NEXT: store i64 0, ptr [[TMP32]], align 8 3559 // CHECK35-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 3560 // CHECK35-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP33]], align 4 3561 // CHECK35-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 3562 // CHECK35-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP34]], align 4 3563 // CHECK35-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 3564 // CHECK35-NEXT: store i32 0, ptr [[TMP35]], align 4 3565 // CHECK35-NEXT: [[TMP36:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, ptr [[KERNEL_ARGS]]) 3566 // CHECK35-NEXT: [[TMP37:%.*]] = icmp ne i32 [[TMP36]], 0 3567 // CHECK35-NEXT: br i1 [[TMP37]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 3568 // CHECK35: omp_offload.failed: 3569 // CHECK35-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i32 [[TMP3]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3:[0-9]+]] 3570 // CHECK35-NEXT: br label [[OMP_OFFLOAD_CONT]] 3571 // CHECK35: omp_offload.cont: 3572 // CHECK35-NEXT: [[TMP38:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 3573 // CHECK35-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10EEiT_(i32 noundef [[TMP38]]) 3574 // CHECK35-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 3575 // CHECK35-NEXT: [[TMP39:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4 3576 // CHECK35-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP39]]) 3577 // CHECK35-NEXT: [[TMP40:%.*]] = load i32, ptr [[RETVAL]], align 4 3578 // CHECK35-NEXT: ret i32 [[TMP40]] 3579 // 3580 // 3581 // CHECK35-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192 3582 // CHECK35-SAME: (i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 3583 // CHECK35-NEXT: entry: 3584 // CHECK35-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 3585 // CHECK35-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 3586 // CHECK35-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3587 // CHECK35-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 3588 // CHECK35-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 3589 // CHECK35-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3590 // CHECK35-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 3591 // CHECK35-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3592 // CHECK35-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 3, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]]) 3593 // CHECK35-NEXT: ret void 3594 // 3595 // 3596 // CHECK35-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.omp_outlined 3597 // CHECK35-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2]] { 3598 // CHECK35-NEXT: entry: 3599 // CHECK35-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 3600 // CHECK35-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 3601 // CHECK35-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4 3602 // CHECK35-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 3603 // CHECK35-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3604 // CHECK35-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3605 // CHECK35-NEXT: [[TMP:%.*]] = alloca i32, align 4 3606 // CHECK35-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3607 // CHECK35-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3608 // CHECK35-NEXT: [[I:%.*]] = alloca i32, align 4 3609 // CHECK35-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3610 // CHECK35-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3611 // CHECK35-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3612 // CHECK35-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3613 // CHECK35-NEXT: [[I3:%.*]] = alloca i32, align 4 3614 // CHECK35-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 3615 // CHECK35-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 3616 // CHECK35-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4 3617 // CHECK35-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 3618 // CHECK35-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3619 // CHECK35-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4 3620 // CHECK35-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 3621 // CHECK35-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3622 // CHECK35-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 3623 // CHECK35-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 3624 // CHECK35-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3625 // CHECK35-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 3626 // CHECK35-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3627 // CHECK35-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3628 // CHECK35-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 3629 // CHECK35-NEXT: store i32 0, ptr [[I]], align 4 3630 // CHECK35-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3631 // CHECK35-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 3632 // CHECK35-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 3633 // CHECK35: omp.precond.then: 3634 // CHECK35-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 3635 // CHECK35-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3636 // CHECK35-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 3637 // CHECK35-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 3638 // CHECK35-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 3639 // CHECK35-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3640 // CHECK35-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 3641 // CHECK35-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 3642 // CHECK35-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3643 // CHECK35-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3644 // CHECK35-NEXT: [[CMP4:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 3645 // CHECK35-NEXT: br i1 [[CMP4]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3646 // CHECK35: cond.true: 3647 // CHECK35-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3648 // CHECK35-NEXT: br label [[COND_END:%.*]] 3649 // CHECK35: cond.false: 3650 // CHECK35-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3651 // CHECK35-NEXT: br label [[COND_END]] 3652 // CHECK35: cond.end: 3653 // CHECK35-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 3654 // CHECK35-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 3655 // CHECK35-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 3656 // CHECK35-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 3657 // CHECK35-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3658 // CHECK35: omp.inner.for.cond: 3659 // CHECK35-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6:![0-9]+]] 3660 // CHECK35-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP6]] 3661 // CHECK35-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 3662 // CHECK35-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3663 // CHECK35: omp.inner.for.body: 3664 // CHECK35-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]] 3665 // CHECK35-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP16]], 1 3666 // CHECK35-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3667 // CHECK35-NEXT: store i32 [[ADD]], ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP6]] 3668 // CHECK35-NEXT: [[TMP17:%.*]] = load i32, ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP6]] 3669 // CHECK35-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP17]] 3670 // CHECK35-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP6]] 3671 // CHECK35-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3672 // CHECK35: omp.body.continue: 3673 // CHECK35-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3674 // CHECK35: omp.inner.for.inc: 3675 // CHECK35-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]] 3676 // CHECK35-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP18]], 1 3677 // CHECK35-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]] 3678 // CHECK35-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] 3679 // CHECK35: omp.inner.for.end: 3680 // CHECK35-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3681 // CHECK35: omp.loop.exit: 3682 // CHECK35-NEXT: [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3683 // CHECK35-NEXT: [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4 3684 // CHECK35-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP20]]) 3685 // CHECK35-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 3686 // CHECK35-NEXT: [[TMP22:%.*]] = icmp ne i32 [[TMP21]], 0 3687 // CHECK35-NEXT: br i1 [[TMP22]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 3688 // CHECK35: .omp.final.then: 3689 // CHECK35-NEXT: [[TMP23:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3690 // CHECK35-NEXT: [[SUB7:%.*]] = sub nsw i32 [[TMP23]], 0 3691 // CHECK35-NEXT: [[DIV8:%.*]] = sdiv i32 [[SUB7]], 1 3692 // CHECK35-NEXT: [[MUL9:%.*]] = mul nsw i32 [[DIV8]], 1 3693 // CHECK35-NEXT: [[ADD10:%.*]] = add nsw i32 0, [[MUL9]] 3694 // CHECK35-NEXT: store i32 [[ADD10]], ptr [[I3]], align 4 3695 // CHECK35-NEXT: br label [[DOTOMP_FINAL_DONE]] 3696 // CHECK35: .omp.final.done: 3697 // CHECK35-NEXT: br label [[OMP_PRECOND_END]] 3698 // CHECK35: omp.precond.end: 3699 // CHECK35-NEXT: ret void 3700 // 3701 // 3702 // CHECK35-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 3703 // CHECK35-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 3704 // CHECK35-NEXT: entry: 3705 // CHECK35-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 3706 // CHECK35-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 3707 // CHECK35-NEXT: [[TE:%.*]] = alloca i32, align 4 3708 // CHECK35-NEXT: [[TH:%.*]] = alloca i32, align 4 3709 // CHECK35-NEXT: [[TE_CASTED:%.*]] = alloca i32, align 4 3710 // CHECK35-NEXT: [[TH_CASTED:%.*]] = alloca i32, align 4 3711 // CHECK35-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4 3712 // CHECK35-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4 3713 // CHECK35-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4 3714 // CHECK35-NEXT: [[TMP:%.*]] = alloca i32, align 4 3715 // CHECK35-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 3716 // CHECK35-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 3717 // CHECK35-NEXT: store i32 0, ptr [[TE]], align 4 3718 // CHECK35-NEXT: store i32 128, ptr [[TH]], align 4 3719 // CHECK35-NEXT: [[TMP0:%.*]] = load i32, ptr [[TE]], align 4 3720 // CHECK35-NEXT: store i32 [[TMP0]], ptr [[TE_CASTED]], align 4 3721 // CHECK35-NEXT: [[TMP1:%.*]] = load i32, ptr [[TE_CASTED]], align 4 3722 // CHECK35-NEXT: [[TMP2:%.*]] = load i32, ptr [[TH]], align 4 3723 // CHECK35-NEXT: store i32 [[TMP2]], ptr [[TH_CASTED]], align 4 3724 // CHECK35-NEXT: [[TMP3:%.*]] = load i32, ptr [[TH_CASTED]], align 4 3725 // CHECK35-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3726 // CHECK35-NEXT: store i32 [[TMP1]], ptr [[TMP4]], align 4 3727 // CHECK35-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3728 // CHECK35-NEXT: store i32 [[TMP1]], ptr [[TMP5]], align 4 3729 // CHECK35-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 3730 // CHECK35-NEXT: store ptr null, ptr [[TMP6]], align 4 3731 // CHECK35-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 3732 // CHECK35-NEXT: store i32 [[TMP3]], ptr [[TMP7]], align 4 3733 // CHECK35-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 3734 // CHECK35-NEXT: store i32 [[TMP3]], ptr [[TMP8]], align 4 3735 // CHECK35-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 3736 // CHECK35-NEXT: store ptr null, ptr [[TMP9]], align 4 3737 // CHECK35-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 3738 // CHECK35-NEXT: store ptr [[A]], ptr [[TMP10]], align 4 3739 // CHECK35-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 3740 // CHECK35-NEXT: store ptr [[A]], ptr [[TMP11]], align 4 3741 // CHECK35-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 3742 // CHECK35-NEXT: store ptr null, ptr [[TMP12]], align 4 3743 // CHECK35-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3744 // CHECK35-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3745 // CHECK35-NEXT: [[TMP15:%.*]] = load i32, ptr [[TE]], align 4 3746 // CHECK35-NEXT: [[TMP16:%.*]] = load i32, ptr [[TH]], align 4 3747 // CHECK35-NEXT: [[TMP17:%.*]] = icmp ult i32 [[TMP16]], 1 3748 // CHECK35-NEXT: [[TMP18:%.*]] = select i1 [[TMP17]], i32 [[TMP16]], i32 1 3749 // CHECK35-NEXT: [[TMP19:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP15]], 0 3750 // CHECK35-NEXT: [[TMP20:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP18]], 0 3751 // CHECK35-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 3752 // CHECK35-NEXT: store i32 3, ptr [[TMP21]], align 4 3753 // CHECK35-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 3754 // CHECK35-NEXT: store i32 3, ptr [[TMP22]], align 4 3755 // CHECK35-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 3756 // CHECK35-NEXT: store ptr [[TMP13]], ptr [[TMP23]], align 4 3757 // CHECK35-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 3758 // CHECK35-NEXT: store ptr [[TMP14]], ptr [[TMP24]], align 4 3759 // CHECK35-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 3760 // CHECK35-NEXT: store ptr @.offload_sizes.1, ptr [[TMP25]], align 4 3761 // CHECK35-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 3762 // CHECK35-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP26]], align 4 3763 // CHECK35-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 3764 // CHECK35-NEXT: store ptr null, ptr [[TMP27]], align 4 3765 // CHECK35-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 3766 // CHECK35-NEXT: store ptr null, ptr [[TMP28]], align 4 3767 // CHECK35-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 3768 // CHECK35-NEXT: store i64 10, ptr [[TMP29]], align 8 3769 // CHECK35-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 3770 // CHECK35-NEXT: store i64 0, ptr [[TMP30]], align 8 3771 // CHECK35-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 3772 // CHECK35-NEXT: store [3 x i32] [[TMP19]], ptr [[TMP31]], align 4 3773 // CHECK35-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 3774 // CHECK35-NEXT: store [3 x i32] [[TMP20]], ptr [[TMP32]], align 4 3775 // CHECK35-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 3776 // CHECK35-NEXT: store i32 0, ptr [[TMP33]], align 4 3777 // CHECK35-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 [[TMP15]], i32 [[TMP18]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, ptr [[KERNEL_ARGS]]) 3778 // CHECK35-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0 3779 // CHECK35-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 3780 // CHECK35: omp_offload.failed: 3781 // CHECK35-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i32 [[TMP1]], i32 [[TMP3]], ptr [[A]]) #[[ATTR3]] 3782 // CHECK35-NEXT: br label [[OMP_OFFLOAD_CONT]] 3783 // CHECK35: omp_offload.cont: 3784 // CHECK35-NEXT: ret i32 0 3785 // 3786 // 3787 // CHECK35-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181 3788 // CHECK35-SAME: (i32 noundef [[TE:%.*]], i32 noundef [[TH:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 3789 // CHECK35-NEXT: entry: 3790 // CHECK35-NEXT: [[TE_ADDR:%.*]] = alloca i32, align 4 3791 // CHECK35-NEXT: [[TH_ADDR:%.*]] = alloca i32, align 4 3792 // CHECK35-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3793 // CHECK35-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2]]) 3794 // CHECK35-NEXT: store i32 [[TE]], ptr [[TE_ADDR]], align 4 3795 // CHECK35-NEXT: store i32 [[TH]], ptr [[TH_ADDR]], align 4 3796 // CHECK35-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3797 // CHECK35-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3798 // CHECK35-NEXT: [[TMP2:%.*]] = load i32, ptr [[TE_ADDR]], align 4 3799 // CHECK35-NEXT: [[TMP3:%.*]] = load i32, ptr [[TH_ADDR]], align 4 3800 // CHECK35-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB2]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 3801 // CHECK35-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.omp_outlined, ptr [[TMP1]]) 3802 // CHECK35-NEXT: ret void 3803 // 3804 // 3805 // CHECK35-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.omp_outlined 3806 // CHECK35-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 3807 // CHECK35-NEXT: entry: 3808 // CHECK35-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 3809 // CHECK35-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 3810 // CHECK35-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 3811 // CHECK35-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 3812 // CHECK35-NEXT: [[TMP:%.*]] = alloca i32, align 4 3813 // CHECK35-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 3814 // CHECK35-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 3815 // CHECK35-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 3816 // CHECK35-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 3817 // CHECK35-NEXT: [[I:%.*]] = alloca i32, align 4 3818 // CHECK35-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 3819 // CHECK35-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 3820 // CHECK35-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 3821 // CHECK35-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 3822 // CHECK35-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 3823 // CHECK35-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 3824 // CHECK35-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 3825 // CHECK35-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 3826 // CHECK35-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 3827 // CHECK35-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 3828 // CHECK35-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 3829 // CHECK35-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3830 // CHECK35-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 3831 // CHECK35-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 3832 // CHECK35: cond.true: 3833 // CHECK35-NEXT: br label [[COND_END:%.*]] 3834 // CHECK35: cond.false: 3835 // CHECK35-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 3836 // CHECK35-NEXT: br label [[COND_END]] 3837 // CHECK35: cond.end: 3838 // CHECK35-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 3839 // CHECK35-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 3840 // CHECK35-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 3841 // CHECK35-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 3842 // CHECK35-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 3843 // CHECK35: omp.inner.for.cond: 3844 // CHECK35-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12:![0-9]+]] 3845 // CHECK35-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP12]] 3846 // CHECK35-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 3847 // CHECK35-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 3848 // CHECK35: omp.inner.for.body: 3849 // CHECK35-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]] 3850 // CHECK35-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 3851 // CHECK35-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 3852 // CHECK35-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP12]] 3853 // CHECK35-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP12]] 3854 // CHECK35-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP9]] 3855 // CHECK35-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP12]] 3856 // CHECK35-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 3857 // CHECK35: omp.body.continue: 3858 // CHECK35-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 3859 // CHECK35: omp.inner.for.inc: 3860 // CHECK35-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]] 3861 // CHECK35-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 3862 // CHECK35-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP12]] 3863 // CHECK35-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP13:![0-9]+]] 3864 // CHECK35: omp.inner.for.end: 3865 // CHECK35-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 3866 // CHECK35: omp.loop.exit: 3867 // CHECK35-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 3868 // CHECK35-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 3869 // CHECK35-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0 3870 // CHECK35-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 3871 // CHECK35: .omp.final.then: 3872 // CHECK35-NEXT: store i32 10, ptr [[I]], align 4 3873 // CHECK35-NEXT: br label [[DOTOMP_FINAL_DONE]] 3874 // CHECK35: .omp.final.done: 3875 // CHECK35-NEXT: ret void 3876 // 3877 // 3878 // CHECK37-LABEL: define {{[^@]+}}@main 3879 // CHECK37-SAME: (i32 noundef signext [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 3880 // CHECK37-NEXT: entry: 3881 // CHECK37-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 3882 // CHECK37-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 3883 // CHECK37-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 8 3884 // CHECK37-NEXT: [[N:%.*]] = alloca i32, align 4 3885 // CHECK37-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 8 3886 // CHECK37-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 3887 // CHECK37-NEXT: [[ARGC_CASTED:%.*]] = alloca i64, align 8 3888 // CHECK37-NEXT: [[N_CASTED:%.*]] = alloca i64, align 8 3889 // CHECK37-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x ptr], align 8 3890 // CHECK37-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x ptr], align 8 3891 // CHECK37-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x ptr], align 8 3892 // CHECK37-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [4 x i64], align 8 3893 // CHECK37-NEXT: [[TMP:%.*]] = alloca i32, align 4 3894 // CHECK37-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 3895 // CHECK37-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 3896 // CHECK37-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 3897 // CHECK37-NEXT: store i32 0, ptr [[RETVAL]], align 4 3898 // CHECK37-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 3899 // CHECK37-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8 3900 // CHECK37-NEXT: store i32 100, ptr [[N]], align 4 3901 // CHECK37-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 3902 // CHECK37-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 3903 // CHECK37-NEXT: [[TMP2:%.*]] = call ptr @llvm.stacksave.p0() 3904 // CHECK37-NEXT: store ptr [[TMP2]], ptr [[SAVED_STACK]], align 8 3905 // CHECK37-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 3906 // CHECK37-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8 3907 // CHECK37-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 3908 // CHECK37-NEXT: store i32 [[TMP3]], ptr [[ARGC_CASTED]], align 4 3909 // CHECK37-NEXT: [[TMP4:%.*]] = load i64, ptr [[ARGC_CASTED]], align 8 3910 // CHECK37-NEXT: [[TMP5:%.*]] = load i32, ptr [[N]], align 4 3911 // CHECK37-NEXT: store i32 [[TMP5]], ptr [[N_CASTED]], align 4 3912 // CHECK37-NEXT: [[TMP6:%.*]] = load i64, ptr [[N_CASTED]], align 8 3913 // CHECK37-NEXT: [[TMP7:%.*]] = mul nuw i64 [[TMP1]], 4 3914 // CHECK37-NEXT: call void @llvm.memcpy.p0.p0.i64(ptr align 8 [[DOTOFFLOAD_SIZES]], ptr align 8 @.offload_sizes, i64 32, i1 false) 3915 // CHECK37-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3916 // CHECK37-NEXT: store i64 [[TMP4]], ptr [[TMP8]], align 8 3917 // CHECK37-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3918 // CHECK37-NEXT: store i64 [[TMP4]], ptr [[TMP9]], align 8 3919 // CHECK37-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 3920 // CHECK37-NEXT: store ptr null, ptr [[TMP10]], align 8 3921 // CHECK37-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 3922 // CHECK37-NEXT: store i64 [[TMP6]], ptr [[TMP11]], align 8 3923 // CHECK37-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 3924 // CHECK37-NEXT: store i64 [[TMP6]], ptr [[TMP12]], align 8 3925 // CHECK37-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 3926 // CHECK37-NEXT: store ptr null, ptr [[TMP13]], align 8 3927 // CHECK37-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 3928 // CHECK37-NEXT: store i64 [[TMP1]], ptr [[TMP14]], align 8 3929 // CHECK37-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 3930 // CHECK37-NEXT: store i64 [[TMP1]], ptr [[TMP15]], align 8 3931 // CHECK37-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 3932 // CHECK37-NEXT: store ptr null, ptr [[TMP16]], align 8 3933 // CHECK37-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 3934 // CHECK37-NEXT: store ptr [[VLA]], ptr [[TMP17]], align 8 3935 // CHECK37-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3 3936 // CHECK37-NEXT: store ptr [[VLA]], ptr [[TMP18]], align 8 3937 // CHECK37-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 3 3938 // CHECK37-NEXT: store i64 [[TMP7]], ptr [[TMP19]], align 8 3939 // CHECK37-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 3 3940 // CHECK37-NEXT: store ptr null, ptr [[TMP20]], align 8 3941 // CHECK37-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 3942 // CHECK37-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 3943 // CHECK37-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 3944 // CHECK37-NEXT: [[TMP24:%.*]] = load i32, ptr [[N]], align 4 3945 // CHECK37-NEXT: store i32 [[TMP24]], ptr [[DOTCAPTURE_EXPR_]], align 4 3946 // CHECK37-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 3947 // CHECK37-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP25]], 0 3948 // CHECK37-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 3949 // CHECK37-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 3950 // CHECK37-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 3951 // CHECK37-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 3952 // CHECK37-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP26]], 1 3953 // CHECK37-NEXT: [[TMP27:%.*]] = zext i32 [[ADD]] to i64 3954 // CHECK37-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 3955 // CHECK37-NEXT: store i32 3, ptr [[TMP28]], align 4 3956 // CHECK37-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 3957 // CHECK37-NEXT: store i32 4, ptr [[TMP29]], align 4 3958 // CHECK37-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 3959 // CHECK37-NEXT: store ptr [[TMP21]], ptr [[TMP30]], align 8 3960 // CHECK37-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 3961 // CHECK37-NEXT: store ptr [[TMP22]], ptr [[TMP31]], align 8 3962 // CHECK37-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 3963 // CHECK37-NEXT: store ptr [[TMP23]], ptr [[TMP32]], align 8 3964 // CHECK37-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 3965 // CHECK37-NEXT: store ptr @.offload_maptypes, ptr [[TMP33]], align 8 3966 // CHECK37-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 3967 // CHECK37-NEXT: store ptr null, ptr [[TMP34]], align 8 3968 // CHECK37-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 3969 // CHECK37-NEXT: store ptr null, ptr [[TMP35]], align 8 3970 // CHECK37-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 3971 // CHECK37-NEXT: store i64 [[TMP27]], ptr [[TMP36]], align 8 3972 // CHECK37-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 3973 // CHECK37-NEXT: store i64 0, ptr [[TMP37]], align 8 3974 // CHECK37-NEXT: [[TMP38:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 3975 // CHECK37-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP38]], align 4 3976 // CHECK37-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 3977 // CHECK37-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP39]], align 4 3978 // CHECK37-NEXT: [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 3979 // CHECK37-NEXT: store i32 0, ptr [[TMP40]], align 4 3980 // CHECK37-NEXT: [[TMP41:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, ptr [[KERNEL_ARGS]]) 3981 // CHECK37-NEXT: [[TMP42:%.*]] = icmp ne i32 [[TMP41]], 0 3982 // CHECK37-NEXT: br i1 [[TMP42]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 3983 // CHECK37: omp_offload.failed: 3984 // CHECK37-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i64 [[TMP4]], i64 [[TMP6]], i64 [[TMP1]], ptr [[VLA]]) #[[ATTR3:[0-9]+]] 3985 // CHECK37-NEXT: br label [[OMP_OFFLOAD_CONT]] 3986 // CHECK37: omp_offload.cont: 3987 // CHECK37-NEXT: [[TMP43:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 3988 // CHECK37-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10EEiT_(i32 noundef signext [[TMP43]]) 3989 // CHECK37-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 3990 // CHECK37-NEXT: [[TMP44:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 3991 // CHECK37-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP44]]) 3992 // CHECK37-NEXT: [[TMP45:%.*]] = load i32, ptr [[RETVAL]], align 4 3993 // CHECK37-NEXT: ret i32 [[TMP45]] 3994 // 3995 // 3996 // CHECK37-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192 3997 // CHECK37-SAME: (i64 noundef [[ARGC:%.*]], i64 noundef [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 3998 // CHECK37-NEXT: entry: 3999 // CHECK37-NEXT: [[ARGC_ADDR:%.*]] = alloca i64, align 8 4000 // CHECK37-NEXT: [[N_ADDR:%.*]] = alloca i64, align 8 4001 // CHECK37-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 4002 // CHECK37-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 4003 // CHECK37-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 4004 // CHECK37-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i64, align 8 4005 // CHECK37-NEXT: store i64 [[ARGC]], ptr [[ARGC_ADDR]], align 8 4006 // CHECK37-NEXT: store i64 [[N]], ptr [[N_ADDR]], align 8 4007 // CHECK37-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 4008 // CHECK37-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 4009 // CHECK37-NEXT: [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 4010 // CHECK37-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 4011 // CHECK37-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 4012 // CHECK37-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP2]], 0 4013 // CHECK37-NEXT: [[STOREDV:%.*]] = zext i1 [[TOBOOL]] to i8 4014 // CHECK37-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR_]], align 1 4015 // CHECK37-NEXT: [[TMP3:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1 4016 // CHECK37-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 4017 // CHECK37-NEXT: [[STOREDV1:%.*]] = zext i1 [[LOADEDV]] to i8 4018 // CHECK37-NEXT: store i8 [[STOREDV1]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1 4019 // CHECK37-NEXT: [[TMP4:%.*]] = load i64, ptr [[DOTCAPTURE_EXPR__CASTED]], align 8 4020 // CHECK37-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.omp_outlined, ptr [[N_ADDR]], i64 [[TMP0]], ptr [[TMP1]], i64 [[TMP4]]) 4021 // CHECK37-NEXT: ret void 4022 // 4023 // 4024 // CHECK37-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.omp_outlined 4025 // CHECK37-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i64 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i64 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 4026 // CHECK37-NEXT: entry: 4027 // CHECK37-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 4028 // CHECK37-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 4029 // CHECK37-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 8 4030 // CHECK37-NEXT: [[VLA_ADDR:%.*]] = alloca i64, align 8 4031 // CHECK37-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 4032 // CHECK37-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i64, align 8 4033 // CHECK37-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4034 // CHECK37-NEXT: [[TMP:%.*]] = alloca i32, align 4 4035 // CHECK37-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 4036 // CHECK37-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 4037 // CHECK37-NEXT: [[I:%.*]] = alloca i32, align 4 4038 // CHECK37-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4039 // CHECK37-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4040 // CHECK37-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 4041 // CHECK37-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 4042 // CHECK37-NEXT: [[I4:%.*]] = alloca i32, align 4 4043 // CHECK37-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 4044 // CHECK37-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 4045 // CHECK37-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 8 4046 // CHECK37-NEXT: store i64 [[VLA]], ptr [[VLA_ADDR]], align 8 4047 // CHECK37-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 4048 // CHECK37-NEXT: store i64 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 8 4049 // CHECK37-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 8 4050 // CHECK37-NEXT: [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR]], align 8 4051 // CHECK37-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 8 4052 // CHECK37-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 4053 // CHECK37-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4 4054 // CHECK37-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 4055 // CHECK37-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 4056 // CHECK37-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 4057 // CHECK37-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 4058 // CHECK37-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4 4059 // CHECK37-NEXT: store i32 0, ptr [[I]], align 4 4060 // CHECK37-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 4061 // CHECK37-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 4062 // CHECK37-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 4063 // CHECK37: omp.precond.then: 4064 // CHECK37-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 4065 // CHECK37-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 4066 // CHECK37-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 4067 // CHECK37-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 4068 // CHECK37-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 4069 // CHECK37-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 4070 // CHECK37-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 4071 // CHECK37-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 4072 // CHECK37-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4073 // CHECK37-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 4074 // CHECK37-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 4075 // CHECK37-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 4076 // CHECK37: cond.true: 4077 // CHECK37-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 4078 // CHECK37-NEXT: br label [[COND_END:%.*]] 4079 // CHECK37: cond.false: 4080 // CHECK37-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4081 // CHECK37-NEXT: br label [[COND_END]] 4082 // CHECK37: cond.end: 4083 // CHECK37-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 4084 // CHECK37-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 4085 // CHECK37-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 4086 // CHECK37-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 4087 // CHECK37-NEXT: [[TMP14:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1 4088 // CHECK37-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP14]] to i1 4089 // CHECK37-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] 4090 // CHECK37: omp_if.then: 4091 // CHECK37-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4092 // CHECK37: omp.inner.for.cond: 4093 // CHECK37-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5:![0-9]+]] 4094 // CHECK37-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP5]] 4095 // CHECK37-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 4096 // CHECK37-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4097 // CHECK37: omp.inner.for.body: 4098 // CHECK37-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 4099 // CHECK37-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 4100 // CHECK37-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4101 // CHECK37-NEXT: store i32 [[ADD]], ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP5]] 4102 // CHECK37-NEXT: [[TMP18:%.*]] = load i32, ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP5]] 4103 // CHECK37-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP18]] to i64 4104 // CHECK37-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM]] 4105 // CHECK37-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP5]] 4106 // CHECK37-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4107 // CHECK37: omp.body.continue: 4108 // CHECK37-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4109 // CHECK37: omp.inner.for.inc: 4110 // CHECK37-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 4111 // CHECK37-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP19]], 1 4112 // CHECK37-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP5]] 4113 // CHECK37-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP6:![0-9]+]] 4114 // CHECK37: omp.inner.for.end: 4115 // CHECK37-NEXT: br label [[OMP_IF_END:%.*]] 4116 // CHECK37: omp_if.else: 4117 // CHECK37-NEXT: br label [[OMP_INNER_FOR_COND8:%.*]] 4118 // CHECK37: omp.inner.for.cond8: 4119 // CHECK37-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4120 // CHECK37-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4121 // CHECK37-NEXT: [[CMP9:%.*]] = icmp sle i32 [[TMP20]], [[TMP21]] 4122 // CHECK37-NEXT: br i1 [[CMP9]], label [[OMP_INNER_FOR_BODY10:%.*]], label [[OMP_INNER_FOR_END18:%.*]] 4123 // CHECK37: omp.inner.for.body10: 4124 // CHECK37-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4125 // CHECK37-NEXT: [[MUL11:%.*]] = mul nsw i32 [[TMP22]], 1 4126 // CHECK37-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]] 4127 // CHECK37-NEXT: store i32 [[ADD12]], ptr [[I4]], align 4 4128 // CHECK37-NEXT: [[TMP23:%.*]] = load i32, ptr [[I4]], align 4 4129 // CHECK37-NEXT: [[IDXPROM13:%.*]] = sext i32 [[TMP23]] to i64 4130 // CHECK37-NEXT: [[ARRAYIDX14:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[IDXPROM13]] 4131 // CHECK37-NEXT: store i32 0, ptr [[ARRAYIDX14]], align 4 4132 // CHECK37-NEXT: br label [[OMP_BODY_CONTINUE15:%.*]] 4133 // CHECK37: omp.body.continue15: 4134 // CHECK37-NEXT: br label [[OMP_INNER_FOR_INC16:%.*]] 4135 // CHECK37: omp.inner.for.inc16: 4136 // CHECK37-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4137 // CHECK37-NEXT: [[ADD17:%.*]] = add nsw i32 [[TMP24]], 1 4138 // CHECK37-NEXT: store i32 [[ADD17]], ptr [[DOTOMP_IV]], align 4 4139 // CHECK37-NEXT: br label [[OMP_INNER_FOR_COND8]], !llvm.loop [[LOOP9:![0-9]+]] 4140 // CHECK37: omp.inner.for.end18: 4141 // CHECK37-NEXT: br label [[OMP_IF_END]] 4142 // CHECK37: omp_if.end: 4143 // CHECK37-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 4144 // CHECK37: omp.loop.exit: 4145 // CHECK37-NEXT: [[TMP25:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 4146 // CHECK37-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4 4147 // CHECK37-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP26]]) 4148 // CHECK37-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 4149 // CHECK37-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0 4150 // CHECK37-NEXT: br i1 [[TMP28]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 4151 // CHECK37: .omp.final.then: 4152 // CHECK37-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 4153 // CHECK37-NEXT: [[SUB19:%.*]] = sub nsw i32 [[TMP29]], 0 4154 // CHECK37-NEXT: [[DIV20:%.*]] = sdiv i32 [[SUB19]], 1 4155 // CHECK37-NEXT: [[MUL21:%.*]] = mul nsw i32 [[DIV20]], 1 4156 // CHECK37-NEXT: [[ADD22:%.*]] = add nsw i32 0, [[MUL21]] 4157 // CHECK37-NEXT: store i32 [[ADD22]], ptr [[I4]], align 4 4158 // CHECK37-NEXT: br label [[DOTOMP_FINAL_DONE]] 4159 // CHECK37: .omp.final.done: 4160 // CHECK37-NEXT: br label [[OMP_PRECOND_END]] 4161 // CHECK37: omp.precond.end: 4162 // CHECK37-NEXT: ret void 4163 // 4164 // 4165 // CHECK37-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 4166 // CHECK37-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 4167 // CHECK37-NEXT: entry: 4168 // CHECK37-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 4169 // CHECK37-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 4170 // CHECK37-NEXT: [[TE:%.*]] = alloca i32, align 4 4171 // CHECK37-NEXT: [[TH:%.*]] = alloca i32, align 4 4172 // CHECK37-NEXT: [[TE_CASTED:%.*]] = alloca i64, align 8 4173 // CHECK37-NEXT: [[TH_CASTED:%.*]] = alloca i64, align 8 4174 // CHECK37-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 8 4175 // CHECK37-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 8 4176 // CHECK37-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 8 4177 // CHECK37-NEXT: [[TMP:%.*]] = alloca i32, align 4 4178 // CHECK37-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 4179 // CHECK37-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 4180 // CHECK37-NEXT: store i32 0, ptr [[TE]], align 4 4181 // CHECK37-NEXT: store i32 128, ptr [[TH]], align 4 4182 // CHECK37-NEXT: [[TMP0:%.*]] = load i32, ptr [[TE]], align 4 4183 // CHECK37-NEXT: store i32 [[TMP0]], ptr [[TE_CASTED]], align 4 4184 // CHECK37-NEXT: [[TMP1:%.*]] = load i64, ptr [[TE_CASTED]], align 8 4185 // CHECK37-NEXT: [[TMP2:%.*]] = load i32, ptr [[TH]], align 4 4186 // CHECK37-NEXT: store i32 [[TMP2]], ptr [[TH_CASTED]], align 4 4187 // CHECK37-NEXT: [[TMP3:%.*]] = load i64, ptr [[TH_CASTED]], align 8 4188 // CHECK37-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 4189 // CHECK37-NEXT: store i64 [[TMP1]], ptr [[TMP4]], align 8 4190 // CHECK37-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 4191 // CHECK37-NEXT: store i64 [[TMP1]], ptr [[TMP5]], align 8 4192 // CHECK37-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 0 4193 // CHECK37-NEXT: store ptr null, ptr [[TMP6]], align 8 4194 // CHECK37-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 4195 // CHECK37-NEXT: store i64 [[TMP3]], ptr [[TMP7]], align 8 4196 // CHECK37-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 4197 // CHECK37-NEXT: store i64 [[TMP3]], ptr [[TMP8]], align 8 4198 // CHECK37-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 1 4199 // CHECK37-NEXT: store ptr null, ptr [[TMP9]], align 8 4200 // CHECK37-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 4201 // CHECK37-NEXT: store ptr [[A]], ptr [[TMP10]], align 8 4202 // CHECK37-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 4203 // CHECK37-NEXT: store ptr [[A]], ptr [[TMP11]], align 8 4204 // CHECK37-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i64 0, i64 2 4205 // CHECK37-NEXT: store ptr null, ptr [[TMP12]], align 8 4206 // CHECK37-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 4207 // CHECK37-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 4208 // CHECK37-NEXT: [[TMP15:%.*]] = load i32, ptr [[TE]], align 4 4209 // CHECK37-NEXT: [[TMP16:%.*]] = load i32, ptr [[TH]], align 4 4210 // CHECK37-NEXT: [[TMP17:%.*]] = icmp ult i32 [[TMP16]], 1 4211 // CHECK37-NEXT: [[TMP18:%.*]] = select i1 [[TMP17]], i32 [[TMP16]], i32 1 4212 // CHECK37-NEXT: [[TMP19:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP15]], 0 4213 // CHECK37-NEXT: [[TMP20:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP18]], 0 4214 // CHECK37-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 4215 // CHECK37-NEXT: store i32 3, ptr [[TMP21]], align 4 4216 // CHECK37-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 4217 // CHECK37-NEXT: store i32 3, ptr [[TMP22]], align 4 4218 // CHECK37-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 4219 // CHECK37-NEXT: store ptr [[TMP13]], ptr [[TMP23]], align 8 4220 // CHECK37-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 4221 // CHECK37-NEXT: store ptr [[TMP14]], ptr [[TMP24]], align 8 4222 // CHECK37-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 4223 // CHECK37-NEXT: store ptr @.offload_sizes.1, ptr [[TMP25]], align 8 4224 // CHECK37-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 4225 // CHECK37-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP26]], align 8 4226 // CHECK37-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 4227 // CHECK37-NEXT: store ptr null, ptr [[TMP27]], align 8 4228 // CHECK37-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 4229 // CHECK37-NEXT: store ptr null, ptr [[TMP28]], align 8 4230 // CHECK37-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 4231 // CHECK37-NEXT: store i64 10, ptr [[TMP29]], align 8 4232 // CHECK37-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 4233 // CHECK37-NEXT: store i64 0, ptr [[TMP30]], align 8 4234 // CHECK37-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 4235 // CHECK37-NEXT: store [3 x i32] [[TMP19]], ptr [[TMP31]], align 4 4236 // CHECK37-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 4237 // CHECK37-NEXT: store [3 x i32] [[TMP20]], ptr [[TMP32]], align 4 4238 // CHECK37-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 4239 // CHECK37-NEXT: store i32 0, ptr [[TMP33]], align 4 4240 // CHECK37-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 [[TMP15]], i32 [[TMP18]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, ptr [[KERNEL_ARGS]]) 4241 // CHECK37-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0 4242 // CHECK37-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 4243 // CHECK37: omp_offload.failed: 4244 // CHECK37-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i64 [[TMP1]], i64 [[TMP3]], ptr [[A]]) #[[ATTR3]] 4245 // CHECK37-NEXT: br label [[OMP_OFFLOAD_CONT]] 4246 // CHECK37: omp_offload.cont: 4247 // CHECK37-NEXT: ret i32 0 4248 // 4249 // 4250 // CHECK37-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181 4251 // CHECK37-SAME: (i64 noundef [[TE:%.*]], i64 noundef [[TH:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 4252 // CHECK37-NEXT: entry: 4253 // CHECK37-NEXT: [[TE_ADDR:%.*]] = alloca i64, align 8 4254 // CHECK37-NEXT: [[TH_ADDR:%.*]] = alloca i64, align 8 4255 // CHECK37-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 4256 // CHECK37-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2]]) 4257 // CHECK37-NEXT: store i64 [[TE]], ptr [[TE_ADDR]], align 8 4258 // CHECK37-NEXT: store i64 [[TH]], ptr [[TH_ADDR]], align 8 4259 // CHECK37-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 4260 // CHECK37-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 8 4261 // CHECK37-NEXT: [[TMP2:%.*]] = load i32, ptr [[TE_ADDR]], align 4 4262 // CHECK37-NEXT: [[TMP3:%.*]] = load i32, ptr [[TH_ADDR]], align 4 4263 // CHECK37-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB2]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 4264 // CHECK37-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.omp_outlined, ptr [[TMP1]]) 4265 // CHECK37-NEXT: ret void 4266 // 4267 // 4268 // CHECK37-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.omp_outlined 4269 // CHECK37-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 4270 // CHECK37-NEXT: entry: 4271 // CHECK37-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8 4272 // CHECK37-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8 4273 // CHECK37-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 8 4274 // CHECK37-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4275 // CHECK37-NEXT: [[TMP:%.*]] = alloca i32, align 4 4276 // CHECK37-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4277 // CHECK37-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4278 // CHECK37-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 4279 // CHECK37-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 4280 // CHECK37-NEXT: [[I:%.*]] = alloca i32, align 4 4281 // CHECK37-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8 4282 // CHECK37-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8 4283 // CHECK37-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 8 4284 // CHECK37-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 8 4285 // CHECK37-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 4286 // CHECK37-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 4287 // CHECK37-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 4288 // CHECK37-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 4289 // CHECK37-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8 4290 // CHECK37-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 4291 // CHECK37-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 4292 // CHECK37-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4293 // CHECK37-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 4294 // CHECK37-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 4295 // CHECK37: cond.true: 4296 // CHECK37-NEXT: br label [[COND_END:%.*]] 4297 // CHECK37: cond.false: 4298 // CHECK37-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4299 // CHECK37-NEXT: br label [[COND_END]] 4300 // CHECK37: cond.end: 4301 // CHECK37-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 4302 // CHECK37-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 4303 // CHECK37-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 4304 // CHECK37-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 4305 // CHECK37-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4306 // CHECK37: omp.inner.for.cond: 4307 // CHECK37-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP13:![0-9]+]] 4308 // CHECK37-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP13]] 4309 // CHECK37-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 4310 // CHECK37-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4311 // CHECK37: omp.inner.for.body: 4312 // CHECK37-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP13]] 4313 // CHECK37-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 4314 // CHECK37-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4315 // CHECK37-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP13]] 4316 // CHECK37-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP13]] 4317 // CHECK37-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP9]] to i64 4318 // CHECK37-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i64 0, i64 [[IDXPROM]] 4319 // CHECK37-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP13]] 4320 // CHECK37-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4321 // CHECK37: omp.body.continue: 4322 // CHECK37-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4323 // CHECK37: omp.inner.for.inc: 4324 // CHECK37-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP13]] 4325 // CHECK37-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 4326 // CHECK37-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP13]] 4327 // CHECK37-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP14:![0-9]+]] 4328 // CHECK37: omp.inner.for.end: 4329 // CHECK37-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 4330 // CHECK37: omp.loop.exit: 4331 // CHECK37-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 4332 // CHECK37-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 4333 // CHECK37-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0 4334 // CHECK37-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 4335 // CHECK37: .omp.final.then: 4336 // CHECK37-NEXT: store i32 10, ptr [[I]], align 4 4337 // CHECK37-NEXT: br label [[DOTOMP_FINAL_DONE]] 4338 // CHECK37: .omp.final.done: 4339 // CHECK37-NEXT: ret void 4340 // 4341 // 4342 // CHECK39-LABEL: define {{[^@]+}}@main 4343 // CHECK39-SAME: (i32 noundef [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 4344 // CHECK39-NEXT: entry: 4345 // CHECK39-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 4346 // CHECK39-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 4347 // CHECK39-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 4 4348 // CHECK39-NEXT: [[N:%.*]] = alloca i32, align 4 4349 // CHECK39-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 4 4350 // CHECK39-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 4351 // CHECK39-NEXT: [[ARGC_CASTED:%.*]] = alloca i32, align 4 4352 // CHECK39-NEXT: [[N_CASTED:%.*]] = alloca i32, align 4 4353 // CHECK39-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [4 x ptr], align 4 4354 // CHECK39-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [4 x ptr], align 4 4355 // CHECK39-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [4 x ptr], align 4 4356 // CHECK39-NEXT: [[DOTOFFLOAD_SIZES:%.*]] = alloca [4 x i64], align 4 4357 // CHECK39-NEXT: [[TMP:%.*]] = alloca i32, align 4 4358 // CHECK39-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 4359 // CHECK39-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 4360 // CHECK39-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 4361 // CHECK39-NEXT: store i32 0, ptr [[RETVAL]], align 4 4362 // CHECK39-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 4363 // CHECK39-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 4 4364 // CHECK39-NEXT: store i32 100, ptr [[N]], align 4 4365 // CHECK39-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 4366 // CHECK39-NEXT: [[TMP1:%.*]] = call ptr @llvm.stacksave.p0() 4367 // CHECK39-NEXT: store ptr [[TMP1]], ptr [[SAVED_STACK]], align 4 4368 // CHECK39-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 4369 // CHECK39-NEXT: store i32 [[TMP0]], ptr [[__VLA_EXPR0]], align 4 4370 // CHECK39-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 4371 // CHECK39-NEXT: store i32 [[TMP2]], ptr [[ARGC_CASTED]], align 4 4372 // CHECK39-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARGC_CASTED]], align 4 4373 // CHECK39-NEXT: [[TMP4:%.*]] = load i32, ptr [[N]], align 4 4374 // CHECK39-NEXT: store i32 [[TMP4]], ptr [[N_CASTED]], align 4 4375 // CHECK39-NEXT: [[TMP5:%.*]] = load i32, ptr [[N_CASTED]], align 4 4376 // CHECK39-NEXT: [[TMP6:%.*]] = mul nuw i32 [[TMP0]], 4 4377 // CHECK39-NEXT: [[TMP7:%.*]] = sext i32 [[TMP6]] to i64 4378 // CHECK39-NEXT: call void @llvm.memcpy.p0.p0.i32(ptr align 4 [[DOTOFFLOAD_SIZES]], ptr align 4 @.offload_sizes, i32 32, i1 false) 4379 // CHECK39-NEXT: [[TMP8:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 4380 // CHECK39-NEXT: store i32 [[TMP3]], ptr [[TMP8]], align 4 4381 // CHECK39-NEXT: [[TMP9:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 4382 // CHECK39-NEXT: store i32 [[TMP3]], ptr [[TMP9]], align 4 4383 // CHECK39-NEXT: [[TMP10:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 4384 // CHECK39-NEXT: store ptr null, ptr [[TMP10]], align 4 4385 // CHECK39-NEXT: [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 4386 // CHECK39-NEXT: store i32 [[TMP5]], ptr [[TMP11]], align 4 4387 // CHECK39-NEXT: [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 4388 // CHECK39-NEXT: store i32 [[TMP5]], ptr [[TMP12]], align 4 4389 // CHECK39-NEXT: [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 4390 // CHECK39-NEXT: store ptr null, ptr [[TMP13]], align 4 4391 // CHECK39-NEXT: [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 4392 // CHECK39-NEXT: store i32 [[TMP0]], ptr [[TMP14]], align 4 4393 // CHECK39-NEXT: [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 4394 // CHECK39-NEXT: store i32 [[TMP0]], ptr [[TMP15]], align 4 4395 // CHECK39-NEXT: [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 4396 // CHECK39-NEXT: store ptr null, ptr [[TMP16]], align 4 4397 // CHECK39-NEXT: [[TMP17:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 3 4398 // CHECK39-NEXT: store ptr [[VLA]], ptr [[TMP17]], align 4 4399 // CHECK39-NEXT: [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 3 4400 // CHECK39-NEXT: store ptr [[VLA]], ptr [[TMP18]], align 4 4401 // CHECK39-NEXT: [[TMP19:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 3 4402 // CHECK39-NEXT: store i64 [[TMP7]], ptr [[TMP19]], align 4 4403 // CHECK39-NEXT: [[TMP20:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 3 4404 // CHECK39-NEXT: store ptr null, ptr [[TMP20]], align 4 4405 // CHECK39-NEXT: [[TMP21:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 4406 // CHECK39-NEXT: [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 4407 // CHECK39-NEXT: [[TMP23:%.*]] = getelementptr inbounds [4 x i64], ptr [[DOTOFFLOAD_SIZES]], i32 0, i32 0 4408 // CHECK39-NEXT: [[TMP24:%.*]] = load i32, ptr [[N]], align 4 4409 // CHECK39-NEXT: store i32 [[TMP24]], ptr [[DOTCAPTURE_EXPR_]], align 4 4410 // CHECK39-NEXT: [[TMP25:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 4411 // CHECK39-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP25]], 0 4412 // CHECK39-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 4413 // CHECK39-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 4414 // CHECK39-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 4415 // CHECK39-NEXT: [[TMP26:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 4416 // CHECK39-NEXT: [[ADD:%.*]] = add nsw i32 [[TMP26]], 1 4417 // CHECK39-NEXT: [[TMP27:%.*]] = zext i32 [[ADD]] to i64 4418 // CHECK39-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 4419 // CHECK39-NEXT: store i32 3, ptr [[TMP28]], align 4 4420 // CHECK39-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 4421 // CHECK39-NEXT: store i32 4, ptr [[TMP29]], align 4 4422 // CHECK39-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 4423 // CHECK39-NEXT: store ptr [[TMP21]], ptr [[TMP30]], align 4 4424 // CHECK39-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 4425 // CHECK39-NEXT: store ptr [[TMP22]], ptr [[TMP31]], align 4 4426 // CHECK39-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 4427 // CHECK39-NEXT: store ptr [[TMP23]], ptr [[TMP32]], align 4 4428 // CHECK39-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 4429 // CHECK39-NEXT: store ptr @.offload_maptypes, ptr [[TMP33]], align 4 4430 // CHECK39-NEXT: [[TMP34:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 4431 // CHECK39-NEXT: store ptr null, ptr [[TMP34]], align 4 4432 // CHECK39-NEXT: [[TMP35:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 4433 // CHECK39-NEXT: store ptr null, ptr [[TMP35]], align 4 4434 // CHECK39-NEXT: [[TMP36:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 4435 // CHECK39-NEXT: store i64 [[TMP27]], ptr [[TMP36]], align 8 4436 // CHECK39-NEXT: [[TMP37:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 4437 // CHECK39-NEXT: store i64 0, ptr [[TMP37]], align 8 4438 // CHECK39-NEXT: [[TMP38:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 4439 // CHECK39-NEXT: store [3 x i32] zeroinitializer, ptr [[TMP38]], align 4 4440 // CHECK39-NEXT: [[TMP39:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 4441 // CHECK39-NEXT: store [3 x i32] [i32 1, i32 0, i32 0], ptr [[TMP39]], align 4 4442 // CHECK39-NEXT: [[TMP40:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 4443 // CHECK39-NEXT: store i32 0, ptr [[TMP40]], align 4 4444 // CHECK39-NEXT: [[TMP41:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2:[0-9]+]], i64 -1, i32 0, i32 1, ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.region_id, ptr [[KERNEL_ARGS]]) 4445 // CHECK39-NEXT: [[TMP42:%.*]] = icmp ne i32 [[TMP41]], 0 4446 // CHECK39-NEXT: br i1 [[TMP42]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 4447 // CHECK39: omp_offload.failed: 4448 // CHECK39-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192(i32 [[TMP3]], i32 [[TMP5]], i32 [[TMP0]], ptr [[VLA]]) #[[ATTR3:[0-9]+]] 4449 // CHECK39-NEXT: br label [[OMP_OFFLOAD_CONT]] 4450 // CHECK39: omp_offload.cont: 4451 // CHECK39-NEXT: [[TMP43:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 4452 // CHECK39-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10EEiT_(i32 noundef [[TMP43]]) 4453 // CHECK39-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 4454 // CHECK39-NEXT: [[TMP44:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4 4455 // CHECK39-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP44]]) 4456 // CHECK39-NEXT: [[TMP45:%.*]] = load i32, ptr [[RETVAL]], align 4 4457 // CHECK39-NEXT: ret i32 [[TMP45]] 4458 // 4459 // 4460 // CHECK39-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192 4461 // CHECK39-SAME: (i32 noundef [[ARGC:%.*]], i32 noundef [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]]) #[[ATTR2:[0-9]+]] { 4462 // CHECK39-NEXT: entry: 4463 // CHECK39-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 4464 // CHECK39-NEXT: [[N_ADDR:%.*]] = alloca i32, align 4 4465 // CHECK39-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 4466 // CHECK39-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 4467 // CHECK39-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 4468 // CHECK39-NEXT: [[DOTCAPTURE_EXPR__CASTED:%.*]] = alloca i32, align 4 4469 // CHECK39-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 4470 // CHECK39-NEXT: store i32 [[N]], ptr [[N_ADDR]], align 4 4471 // CHECK39-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 4472 // CHECK39-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 4473 // CHECK39-NEXT: [[TMP0:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 4474 // CHECK39-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 4475 // CHECK39-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 4476 // CHECK39-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP2]], 0 4477 // CHECK39-NEXT: [[STOREDV:%.*]] = zext i1 [[TOBOOL]] to i8 4478 // CHECK39-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR_]], align 1 4479 // CHECK39-NEXT: [[TMP3:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1 4480 // CHECK39-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP3]] to i1 4481 // CHECK39-NEXT: [[STOREDV1:%.*]] = zext i1 [[LOADEDV]] to i8 4482 // CHECK39-NEXT: store i8 [[STOREDV1]], ptr [[DOTCAPTURE_EXPR__CASTED]], align 1 4483 // CHECK39-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR__CASTED]], align 4 4484 // CHECK39-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 4, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.omp_outlined, ptr [[N_ADDR]], i32 [[TMP0]], ptr [[TMP1]], i32 [[TMP4]]) 4485 // CHECK39-NEXT: ret void 4486 // 4487 // 4488 // CHECK39-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}_main_l192.omp_outlined 4489 // CHECK39-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[N:%.*]], i32 noundef [[VLA:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[A:%.*]], i32 noundef [[DOTCAPTURE_EXPR_:%.*]]) #[[ATTR2]] { 4490 // CHECK39-NEXT: entry: 4491 // CHECK39-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 4492 // CHECK39-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 4493 // CHECK39-NEXT: [[N_ADDR:%.*]] = alloca ptr, align 4 4494 // CHECK39-NEXT: [[VLA_ADDR:%.*]] = alloca i32, align 4 4495 // CHECK39-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 4496 // CHECK39-NEXT: [[DOTCAPTURE_EXPR__ADDR:%.*]] = alloca i32, align 4 4497 // CHECK39-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4498 // CHECK39-NEXT: [[TMP:%.*]] = alloca i32, align 4 4499 // CHECK39-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 4500 // CHECK39-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 4501 // CHECK39-NEXT: [[I:%.*]] = alloca i32, align 4 4502 // CHECK39-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4503 // CHECK39-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4504 // CHECK39-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 4505 // CHECK39-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 4506 // CHECK39-NEXT: [[I4:%.*]] = alloca i32, align 4 4507 // CHECK39-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 4508 // CHECK39-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 4509 // CHECK39-NEXT: store ptr [[N]], ptr [[N_ADDR]], align 4 4510 // CHECK39-NEXT: store i32 [[VLA]], ptr [[VLA_ADDR]], align 4 4511 // CHECK39-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 4512 // CHECK39-NEXT: store i32 [[DOTCAPTURE_EXPR_]], ptr [[DOTCAPTURE_EXPR__ADDR]], align 4 4513 // CHECK39-NEXT: [[TMP0:%.*]] = load ptr, ptr [[N_ADDR]], align 4 4514 // CHECK39-NEXT: [[TMP1:%.*]] = load i32, ptr [[VLA_ADDR]], align 4 4515 // CHECK39-NEXT: [[TMP2:%.*]] = load ptr, ptr [[A_ADDR]], align 4 4516 // CHECK39-NEXT: [[TMP3:%.*]] = load i32, ptr [[TMP0]], align 4 4517 // CHECK39-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4 4518 // CHECK39-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 4519 // CHECK39-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 4520 // CHECK39-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 4521 // CHECK39-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 4522 // CHECK39-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4 4523 // CHECK39-NEXT: store i32 0, ptr [[I]], align 4 4524 // CHECK39-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 4525 // CHECK39-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 4526 // CHECK39-NEXT: br i1 [[CMP]], label [[OMP_PRECOND_THEN:%.*]], label [[OMP_PRECOND_END:%.*]] 4527 // CHECK39: omp.precond.then: 4528 // CHECK39-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 4529 // CHECK39-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 4530 // CHECK39-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 4531 // CHECK39-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 4532 // CHECK39-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 4533 // CHECK39-NEXT: [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 4534 // CHECK39-NEXT: [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4 4535 // CHECK39-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP8]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 4536 // CHECK39-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4537 // CHECK39-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 4538 // CHECK39-NEXT: [[CMP5:%.*]] = icmp sgt i32 [[TMP9]], [[TMP10]] 4539 // CHECK39-NEXT: br i1 [[CMP5]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 4540 // CHECK39: cond.true: 4541 // CHECK39-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 4542 // CHECK39-NEXT: br label [[COND_END:%.*]] 4543 // CHECK39: cond.false: 4544 // CHECK39-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4545 // CHECK39-NEXT: br label [[COND_END]] 4546 // CHECK39: cond.end: 4547 // CHECK39-NEXT: [[COND:%.*]] = phi i32 [ [[TMP11]], [[COND_TRUE]] ], [ [[TMP12]], [[COND_FALSE]] ] 4548 // CHECK39-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 4549 // CHECK39-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 4550 // CHECK39-NEXT: store i32 [[TMP13]], ptr [[DOTOMP_IV]], align 4 4551 // CHECK39-NEXT: [[TMP14:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR__ADDR]], align 1 4552 // CHECK39-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP14]] to i1 4553 // CHECK39-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] 4554 // CHECK39: omp_if.then: 4555 // CHECK39-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4556 // CHECK39: omp.inner.for.cond: 4557 // CHECK39-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6:![0-9]+]] 4558 // CHECK39-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP6]] 4559 // CHECK39-NEXT: [[CMP6:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 4560 // CHECK39-NEXT: br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4561 // CHECK39: omp.inner.for.body: 4562 // CHECK39-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]] 4563 // CHECK39-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1 4564 // CHECK39-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4565 // CHECK39-NEXT: store i32 [[ADD]], ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP6]] 4566 // CHECK39-NEXT: [[TMP18:%.*]] = load i32, ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP6]] 4567 // CHECK39-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP18]] 4568 // CHECK39-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP6]] 4569 // CHECK39-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4570 // CHECK39: omp.body.continue: 4571 // CHECK39-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4572 // CHECK39: omp.inner.for.inc: 4573 // CHECK39-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]] 4574 // CHECK39-NEXT: [[ADD7:%.*]] = add nsw i32 [[TMP19]], 1 4575 // CHECK39-NEXT: store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]] 4576 // CHECK39-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] 4577 // CHECK39: omp.inner.for.end: 4578 // CHECK39-NEXT: br label [[OMP_IF_END:%.*]] 4579 // CHECK39: omp_if.else: 4580 // CHECK39-NEXT: br label [[OMP_INNER_FOR_COND8:%.*]] 4581 // CHECK39: omp.inner.for.cond8: 4582 // CHECK39-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4583 // CHECK39-NEXT: [[TMP21:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4584 // CHECK39-NEXT: [[CMP9:%.*]] = icmp sle i32 [[TMP20]], [[TMP21]] 4585 // CHECK39-NEXT: br i1 [[CMP9]], label [[OMP_INNER_FOR_BODY10:%.*]], label [[OMP_INNER_FOR_END17:%.*]] 4586 // CHECK39: omp.inner.for.body10: 4587 // CHECK39-NEXT: [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4588 // CHECK39-NEXT: [[MUL11:%.*]] = mul nsw i32 [[TMP22]], 1 4589 // CHECK39-NEXT: [[ADD12:%.*]] = add nsw i32 0, [[MUL11]] 4590 // CHECK39-NEXT: store i32 [[ADD12]], ptr [[I4]], align 4 4591 // CHECK39-NEXT: [[TMP23:%.*]] = load i32, ptr [[I4]], align 4 4592 // CHECK39-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i32 [[TMP23]] 4593 // CHECK39-NEXT: store i32 0, ptr [[ARRAYIDX13]], align 4 4594 // CHECK39-NEXT: br label [[OMP_BODY_CONTINUE14:%.*]] 4595 // CHECK39: omp.body.continue14: 4596 // CHECK39-NEXT: br label [[OMP_INNER_FOR_INC15:%.*]] 4597 // CHECK39: omp.inner.for.inc15: 4598 // CHECK39-NEXT: [[TMP24:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 4599 // CHECK39-NEXT: [[ADD16:%.*]] = add nsw i32 [[TMP24]], 1 4600 // CHECK39-NEXT: store i32 [[ADD16]], ptr [[DOTOMP_IV]], align 4 4601 // CHECK39-NEXT: br label [[OMP_INNER_FOR_COND8]], !llvm.loop [[LOOP10:![0-9]+]] 4602 // CHECK39: omp.inner.for.end17: 4603 // CHECK39-NEXT: br label [[OMP_IF_END]] 4604 // CHECK39: omp_if.end: 4605 // CHECK39-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 4606 // CHECK39: omp.loop.exit: 4607 // CHECK39-NEXT: [[TMP25:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 4608 // CHECK39-NEXT: [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4 4609 // CHECK39-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP26]]) 4610 // CHECK39-NEXT: [[TMP27:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 4611 // CHECK39-NEXT: [[TMP28:%.*]] = icmp ne i32 [[TMP27]], 0 4612 // CHECK39-NEXT: br i1 [[TMP28]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 4613 // CHECK39: .omp.final.then: 4614 // CHECK39-NEXT: [[TMP29:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 4615 // CHECK39-NEXT: [[SUB18:%.*]] = sub nsw i32 [[TMP29]], 0 4616 // CHECK39-NEXT: [[DIV19:%.*]] = sdiv i32 [[SUB18]], 1 4617 // CHECK39-NEXT: [[MUL20:%.*]] = mul nsw i32 [[DIV19]], 1 4618 // CHECK39-NEXT: [[ADD21:%.*]] = add nsw i32 0, [[MUL20]] 4619 // CHECK39-NEXT: store i32 [[ADD21]], ptr [[I4]], align 4 4620 // CHECK39-NEXT: br label [[DOTOMP_FINAL_DONE]] 4621 // CHECK39: .omp.final.done: 4622 // CHECK39-NEXT: br label [[OMP_PRECOND_END]] 4623 // CHECK39: omp.precond.end: 4624 // CHECK39-NEXT: ret void 4625 // 4626 // 4627 // CHECK39-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 4628 // CHECK39-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR5:[0-9]+]] comdat { 4629 // CHECK39-NEXT: entry: 4630 // CHECK39-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 4631 // CHECK39-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 4632 // CHECK39-NEXT: [[TE:%.*]] = alloca i32, align 4 4633 // CHECK39-NEXT: [[TH:%.*]] = alloca i32, align 4 4634 // CHECK39-NEXT: [[TE_CASTED:%.*]] = alloca i32, align 4 4635 // CHECK39-NEXT: [[TH_CASTED:%.*]] = alloca i32, align 4 4636 // CHECK39-NEXT: [[DOTOFFLOAD_BASEPTRS:%.*]] = alloca [3 x ptr], align 4 4637 // CHECK39-NEXT: [[DOTOFFLOAD_PTRS:%.*]] = alloca [3 x ptr], align 4 4638 // CHECK39-NEXT: [[DOTOFFLOAD_MAPPERS:%.*]] = alloca [3 x ptr], align 4 4639 // CHECK39-NEXT: [[TMP:%.*]] = alloca i32, align 4 4640 // CHECK39-NEXT: [[KERNEL_ARGS:%.*]] = alloca [[STRUCT___TGT_KERNEL_ARGUMENTS:%.*]], align 8 4641 // CHECK39-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 4642 // CHECK39-NEXT: store i32 0, ptr [[TE]], align 4 4643 // CHECK39-NEXT: store i32 128, ptr [[TH]], align 4 4644 // CHECK39-NEXT: [[TMP0:%.*]] = load i32, ptr [[TE]], align 4 4645 // CHECK39-NEXT: store i32 [[TMP0]], ptr [[TE_CASTED]], align 4 4646 // CHECK39-NEXT: [[TMP1:%.*]] = load i32, ptr [[TE_CASTED]], align 4 4647 // CHECK39-NEXT: [[TMP2:%.*]] = load i32, ptr [[TH]], align 4 4648 // CHECK39-NEXT: store i32 [[TMP2]], ptr [[TH_CASTED]], align 4 4649 // CHECK39-NEXT: [[TMP3:%.*]] = load i32, ptr [[TH_CASTED]], align 4 4650 // CHECK39-NEXT: [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 4651 // CHECK39-NEXT: store i32 [[TMP1]], ptr [[TMP4]], align 4 4652 // CHECK39-NEXT: [[TMP5:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 4653 // CHECK39-NEXT: store i32 [[TMP1]], ptr [[TMP5]], align 4 4654 // CHECK39-NEXT: [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 0 4655 // CHECK39-NEXT: store ptr null, ptr [[TMP6]], align 4 4656 // CHECK39-NEXT: [[TMP7:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 1 4657 // CHECK39-NEXT: store i32 [[TMP3]], ptr [[TMP7]], align 4 4658 // CHECK39-NEXT: [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 1 4659 // CHECK39-NEXT: store i32 [[TMP3]], ptr [[TMP8]], align 4 4660 // CHECK39-NEXT: [[TMP9:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 1 4661 // CHECK39-NEXT: store ptr null, ptr [[TMP9]], align 4 4662 // CHECK39-NEXT: [[TMP10:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 2 4663 // CHECK39-NEXT: store ptr [[A]], ptr [[TMP10]], align 4 4664 // CHECK39-NEXT: [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 2 4665 // CHECK39-NEXT: store ptr [[A]], ptr [[TMP11]], align 4 4666 // CHECK39-NEXT: [[TMP12:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_MAPPERS]], i32 0, i32 2 4667 // CHECK39-NEXT: store ptr null, ptr [[TMP12]], align 4 4668 // CHECK39-NEXT: [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_BASEPTRS]], i32 0, i32 0 4669 // CHECK39-NEXT: [[TMP14:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOFFLOAD_PTRS]], i32 0, i32 0 4670 // CHECK39-NEXT: [[TMP15:%.*]] = load i32, ptr [[TE]], align 4 4671 // CHECK39-NEXT: [[TMP16:%.*]] = load i32, ptr [[TH]], align 4 4672 // CHECK39-NEXT: [[TMP17:%.*]] = icmp ult i32 [[TMP16]], 1 4673 // CHECK39-NEXT: [[TMP18:%.*]] = select i1 [[TMP17]], i32 [[TMP16]], i32 1 4674 // CHECK39-NEXT: [[TMP19:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP15]], 0 4675 // CHECK39-NEXT: [[TMP20:%.*]] = insertvalue [3 x i32] zeroinitializer, i32 [[TMP18]], 0 4676 // CHECK39-NEXT: [[TMP21:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 0 4677 // CHECK39-NEXT: store i32 3, ptr [[TMP21]], align 4 4678 // CHECK39-NEXT: [[TMP22:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 1 4679 // CHECK39-NEXT: store i32 3, ptr [[TMP22]], align 4 4680 // CHECK39-NEXT: [[TMP23:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 2 4681 // CHECK39-NEXT: store ptr [[TMP13]], ptr [[TMP23]], align 4 4682 // CHECK39-NEXT: [[TMP24:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 3 4683 // CHECK39-NEXT: store ptr [[TMP14]], ptr [[TMP24]], align 4 4684 // CHECK39-NEXT: [[TMP25:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 4 4685 // CHECK39-NEXT: store ptr @.offload_sizes.1, ptr [[TMP25]], align 4 4686 // CHECK39-NEXT: [[TMP26:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 5 4687 // CHECK39-NEXT: store ptr @.offload_maptypes.2, ptr [[TMP26]], align 4 4688 // CHECK39-NEXT: [[TMP27:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 6 4689 // CHECK39-NEXT: store ptr null, ptr [[TMP27]], align 4 4690 // CHECK39-NEXT: [[TMP28:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 7 4691 // CHECK39-NEXT: store ptr null, ptr [[TMP28]], align 4 4692 // CHECK39-NEXT: [[TMP29:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 8 4693 // CHECK39-NEXT: store i64 10, ptr [[TMP29]], align 8 4694 // CHECK39-NEXT: [[TMP30:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 9 4695 // CHECK39-NEXT: store i64 0, ptr [[TMP30]], align 8 4696 // CHECK39-NEXT: [[TMP31:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 10 4697 // CHECK39-NEXT: store [3 x i32] [[TMP19]], ptr [[TMP31]], align 4 4698 // CHECK39-NEXT: [[TMP32:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 11 4699 // CHECK39-NEXT: store [3 x i32] [[TMP20]], ptr [[TMP32]], align 4 4700 // CHECK39-NEXT: [[TMP33:%.*]] = getelementptr inbounds nuw [[STRUCT___TGT_KERNEL_ARGUMENTS]], ptr [[KERNEL_ARGS]], i32 0, i32 12 4701 // CHECK39-NEXT: store i32 0, ptr [[TMP33]], align 4 4702 // CHECK39-NEXT: [[TMP34:%.*]] = call i32 @__tgt_target_kernel(ptr @[[GLOB2]], i64 -1, i32 [[TMP15]], i32 [[TMP18]], ptr @.{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.region_id, ptr [[KERNEL_ARGS]]) 4703 // CHECK39-NEXT: [[TMP35:%.*]] = icmp ne i32 [[TMP34]], 0 4704 // CHECK39-NEXT: br i1 [[TMP35]], label [[OMP_OFFLOAD_FAILED:%.*]], label [[OMP_OFFLOAD_CONT:%.*]] 4705 // CHECK39: omp_offload.failed: 4706 // CHECK39-NEXT: call void @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181(i32 [[TMP1]], i32 [[TMP3]], ptr [[A]]) #[[ATTR3]] 4707 // CHECK39-NEXT: br label [[OMP_OFFLOAD_CONT]] 4708 // CHECK39: omp_offload.cont: 4709 // CHECK39-NEXT: ret i32 0 4710 // 4711 // 4712 // CHECK39-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181 4713 // CHECK39-SAME: (i32 noundef [[TE:%.*]], i32 noundef [[TH:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 4714 // CHECK39-NEXT: entry: 4715 // CHECK39-NEXT: [[TE_ADDR:%.*]] = alloca i32, align 4 4716 // CHECK39-NEXT: [[TH_ADDR:%.*]] = alloca i32, align 4 4717 // CHECK39-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 4718 // CHECK39-NEXT: [[TMP0:%.*]] = call i32 @__kmpc_global_thread_num(ptr @[[GLOB2]]) 4719 // CHECK39-NEXT: store i32 [[TE]], ptr [[TE_ADDR]], align 4 4720 // CHECK39-NEXT: store i32 [[TH]], ptr [[TH_ADDR]], align 4 4721 // CHECK39-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 4722 // CHECK39-NEXT: [[TMP1:%.*]] = load ptr, ptr [[A_ADDR]], align 4 4723 // CHECK39-NEXT: [[TMP2:%.*]] = load i32, ptr [[TE_ADDR]], align 4 4724 // CHECK39-NEXT: [[TMP3:%.*]] = load i32, ptr [[TH_ADDR]], align 4 4725 // CHECK39-NEXT: call void @__kmpc_push_num_teams(ptr @[[GLOB2]], i32 [[TMP0]], i32 [[TMP2]], i32 [[TMP3]]) 4726 // CHECK39-NEXT: call void (ptr, i32, ptr, ...) @__kmpc_fork_teams(ptr @[[GLOB2]], i32 1, ptr @{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.omp_outlined, ptr [[TMP1]]) 4727 // CHECK39-NEXT: ret void 4728 // 4729 // 4730 // CHECK39-LABEL: define {{[^@]+}}@{{__omp_offloading_[0-9a-z]+_[0-9a-z]+}}__Z5tmainIiLi10EEiT__l181.omp_outlined 4731 // CHECK39-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(40) [[A:%.*]]) #[[ATTR2]] { 4732 // CHECK39-NEXT: entry: 4733 // CHECK39-NEXT: [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 4 4734 // CHECK39-NEXT: [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 4 4735 // CHECK39-NEXT: [[A_ADDR:%.*]] = alloca ptr, align 4 4736 // CHECK39-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4737 // CHECK39-NEXT: [[TMP:%.*]] = alloca i32, align 4 4738 // CHECK39-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4739 // CHECK39-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4740 // CHECK39-NEXT: [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4 4741 // CHECK39-NEXT: [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4 4742 // CHECK39-NEXT: [[I:%.*]] = alloca i32, align 4 4743 // CHECK39-NEXT: store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 4 4744 // CHECK39-NEXT: store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 4 4745 // CHECK39-NEXT: store ptr [[A]], ptr [[A_ADDR]], align 4 4746 // CHECK39-NEXT: [[TMP0:%.*]] = load ptr, ptr [[A_ADDR]], align 4 4747 // CHECK39-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 4748 // CHECK39-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 4749 // CHECK39-NEXT: store i32 1, ptr [[DOTOMP_STRIDE]], align 4 4750 // CHECK39-NEXT: store i32 0, ptr [[DOTOMP_IS_LAST]], align 4 4751 // CHECK39-NEXT: [[TMP1:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 4 4752 // CHECK39-NEXT: [[TMP2:%.*]] = load i32, ptr [[TMP1]], align 4 4753 // CHECK39-NEXT: call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP2]], i32 92, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1) 4754 // CHECK39-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4755 // CHECK39-NEXT: [[CMP:%.*]] = icmp sgt i32 [[TMP3]], 9 4756 // CHECK39-NEXT: br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]] 4757 // CHECK39: cond.true: 4758 // CHECK39-NEXT: br label [[COND_END:%.*]] 4759 // CHECK39: cond.false: 4760 // CHECK39-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 4761 // CHECK39-NEXT: br label [[COND_END]] 4762 // CHECK39: cond.end: 4763 // CHECK39-NEXT: [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP4]], [[COND_FALSE]] ] 4764 // CHECK39-NEXT: store i32 [[COND]], ptr [[DOTOMP_UB]], align 4 4765 // CHECK39-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 4766 // CHECK39-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_IV]], align 4 4767 // CHECK39-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4768 // CHECK39: omp.inner.for.cond: 4769 // CHECK39-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14:![0-9]+]] 4770 // CHECK39-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP14]] 4771 // CHECK39-NEXT: [[CMP1:%.*]] = icmp sle i32 [[TMP6]], [[TMP7]] 4772 // CHECK39-NEXT: br i1 [[CMP1]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4773 // CHECK39: omp.inner.for.body: 4774 // CHECK39-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14]] 4775 // CHECK39-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP8]], 1 4776 // CHECK39-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4777 // CHECK39-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP14]] 4778 // CHECK39-NEXT: [[TMP9:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP14]] 4779 // CHECK39-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[TMP0]], i32 0, i32 [[TMP9]] 4780 // CHECK39-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP14]] 4781 // CHECK39-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4782 // CHECK39: omp.body.continue: 4783 // CHECK39-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4784 // CHECK39: omp.inner.for.inc: 4785 // CHECK39-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14]] 4786 // CHECK39-NEXT: [[ADD2:%.*]] = add nsw i32 [[TMP10]], 1 4787 // CHECK39-NEXT: store i32 [[ADD2]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP14]] 4788 // CHECK39-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP15:![0-9]+]] 4789 // CHECK39: omp.inner.for.end: 4790 // CHECK39-NEXT: br label [[OMP_LOOP_EXIT:%.*]] 4791 // CHECK39: omp.loop.exit: 4792 // CHECK39-NEXT: call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP2]]) 4793 // CHECK39-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IS_LAST]], align 4 4794 // CHECK39-NEXT: [[TMP12:%.*]] = icmp ne i32 [[TMP11]], 0 4795 // CHECK39-NEXT: br i1 [[TMP12]], label [[DOTOMP_FINAL_THEN:%.*]], label [[DOTOMP_FINAL_DONE:%.*]] 4796 // CHECK39: .omp.final.then: 4797 // CHECK39-NEXT: store i32 10, ptr [[I]], align 4 4798 // CHECK39-NEXT: br label [[DOTOMP_FINAL_DONE]] 4799 // CHECK39: .omp.final.done: 4800 // CHECK39-NEXT: ret void 4801 // 4802 // 4803 // CHECK41-LABEL: define {{[^@]+}}@main 4804 // CHECK41-SAME: (i32 noundef signext [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 4805 // CHECK41-NEXT: entry: 4806 // CHECK41-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 4807 // CHECK41-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 4808 // CHECK41-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 8 4809 // CHECK41-NEXT: [[N:%.*]] = alloca i32, align 4 4810 // CHECK41-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 8 4811 // CHECK41-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 4812 // CHECK41-NEXT: [[TMP:%.*]] = alloca i32, align 4 4813 // CHECK41-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 4814 // CHECK41-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 4815 // CHECK41-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4816 // CHECK41-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4817 // CHECK41-NEXT: [[I:%.*]] = alloca i32, align 4 4818 // CHECK41-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4819 // CHECK41-NEXT: [[I3:%.*]] = alloca i32, align 4 4820 // CHECK41-NEXT: store i32 0, ptr [[RETVAL]], align 4 4821 // CHECK41-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 4822 // CHECK41-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8 4823 // CHECK41-NEXT: store i32 100, ptr [[N]], align 4 4824 // CHECK41-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 4825 // CHECK41-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 4826 // CHECK41-NEXT: [[TMP2:%.*]] = call ptr @llvm.stacksave.p0() 4827 // CHECK41-NEXT: store ptr [[TMP2]], ptr [[SAVED_STACK]], align 8 4828 // CHECK41-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 4829 // CHECK41-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8 4830 // CHECK41-NEXT: [[TMP3:%.*]] = load i32, ptr [[N]], align 4 4831 // CHECK41-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_]], align 4 4832 // CHECK41-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 4833 // CHECK41-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 4834 // CHECK41-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 4835 // CHECK41-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 4836 // CHECK41-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 4837 // CHECK41-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 4838 // CHECK41-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 4839 // CHECK41-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_UB]], align 4 4840 // CHECK41-NEXT: store i32 0, ptr [[I]], align 4 4841 // CHECK41-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 4842 // CHECK41-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 4843 // CHECK41-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 4844 // CHECK41: simd.if.then: 4845 // CHECK41-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 4846 // CHECK41-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 4847 // CHECK41-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4848 // CHECK41: omp.inner.for.cond: 4849 // CHECK41-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2:![0-9]+]] 4850 // CHECK41-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP2]] 4851 // CHECK41-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP8]], [[TMP9]] 4852 // CHECK41-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4853 // CHECK41: omp.inner.for.body: 4854 // CHECK41-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 4855 // CHECK41-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP10]], 1 4856 // CHECK41-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4857 // CHECK41-NEXT: store i32 [[ADD]], ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP2]] 4858 // CHECK41-NEXT: [[TMP11:%.*]] = load i32, ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP2]] 4859 // CHECK41-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP11]] to i64 4860 // CHECK41-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i64 [[IDXPROM]] 4861 // CHECK41-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP2]] 4862 // CHECK41-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4863 // CHECK41: omp.body.continue: 4864 // CHECK41-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4865 // CHECK41: omp.inner.for.inc: 4866 // CHECK41-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 4867 // CHECK41-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP12]], 1 4868 // CHECK41-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 4869 // CHECK41-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] 4870 // CHECK41: omp.inner.for.end: 4871 // CHECK41-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 4872 // CHECK41-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP13]], 0 4873 // CHECK41-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 4874 // CHECK41-NEXT: [[MUL8:%.*]] = mul nsw i32 [[DIV7]], 1 4875 // CHECK41-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL8]] 4876 // CHECK41-NEXT: store i32 [[ADD9]], ptr [[I3]], align 4 4877 // CHECK41-NEXT: br label [[SIMD_IF_END]] 4878 // CHECK41: simd.if.end: 4879 // CHECK41-NEXT: [[TMP14:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 4880 // CHECK41-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10EEiT_(i32 noundef signext [[TMP14]]) 4881 // CHECK41-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 4882 // CHECK41-NEXT: [[TMP15:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 4883 // CHECK41-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP15]]) 4884 // CHECK41-NEXT: [[TMP16:%.*]] = load i32, ptr [[RETVAL]], align 4 4885 // CHECK41-NEXT: ret i32 [[TMP16]] 4886 // 4887 // 4888 // CHECK41-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 4889 // CHECK41-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] comdat { 4890 // CHECK41-NEXT: entry: 4891 // CHECK41-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 4892 // CHECK41-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 4893 // CHECK41-NEXT: [[TE:%.*]] = alloca i32, align 4 4894 // CHECK41-NEXT: [[TH:%.*]] = alloca i32, align 4 4895 // CHECK41-NEXT: [[TMP:%.*]] = alloca i32, align 4 4896 // CHECK41-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4897 // CHECK41-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4898 // CHECK41-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4899 // CHECK41-NEXT: [[I:%.*]] = alloca i32, align 4 4900 // CHECK41-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 4901 // CHECK41-NEXT: store i32 0, ptr [[TE]], align 4 4902 // CHECK41-NEXT: store i32 128, ptr [[TH]], align 4 4903 // CHECK41-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 4904 // CHECK41-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 4905 // CHECK41-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 4906 // CHECK41-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 4907 // CHECK41-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4908 // CHECK41: omp.inner.for.cond: 4909 // CHECK41-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6:![0-9]+]] 4910 // CHECK41-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP6]] 4911 // CHECK41-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] 4912 // CHECK41-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4913 // CHECK41: omp.inner.for.body: 4914 // CHECK41-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]] 4915 // CHECK41-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 4916 // CHECK41-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4917 // CHECK41-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]] 4918 // CHECK41-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP6]] 4919 // CHECK41-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP4]] to i64 4920 // CHECK41-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]] 4921 // CHECK41-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP6]] 4922 // CHECK41-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4923 // CHECK41: omp.body.continue: 4924 // CHECK41-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4925 // CHECK41: omp.inner.for.inc: 4926 // CHECK41-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]] 4927 // CHECK41-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], 1 4928 // CHECK41-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP6]] 4929 // CHECK41-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP7:![0-9]+]] 4930 // CHECK41: omp.inner.for.end: 4931 // CHECK41-NEXT: store i32 10, ptr [[I]], align 4 4932 // CHECK41-NEXT: ret i32 0 4933 // 4934 // 4935 // CHECK43-LABEL: define {{[^@]+}}@main 4936 // CHECK43-SAME: (i32 noundef [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 4937 // CHECK43-NEXT: entry: 4938 // CHECK43-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 4939 // CHECK43-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 4940 // CHECK43-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 4 4941 // CHECK43-NEXT: [[N:%.*]] = alloca i32, align 4 4942 // CHECK43-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 4 4943 // CHECK43-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 4944 // CHECK43-NEXT: [[TMP:%.*]] = alloca i32, align 4 4945 // CHECK43-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i32, align 4 4946 // CHECK43-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 4947 // CHECK43-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 4948 // CHECK43-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 4949 // CHECK43-NEXT: [[I:%.*]] = alloca i32, align 4 4950 // CHECK43-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 4951 // CHECK43-NEXT: [[I3:%.*]] = alloca i32, align 4 4952 // CHECK43-NEXT: store i32 0, ptr [[RETVAL]], align 4 4953 // CHECK43-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 4954 // CHECK43-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 4 4955 // CHECK43-NEXT: store i32 100, ptr [[N]], align 4 4956 // CHECK43-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 4957 // CHECK43-NEXT: [[TMP1:%.*]] = call ptr @llvm.stacksave.p0() 4958 // CHECK43-NEXT: store ptr [[TMP1]], ptr [[SAVED_STACK]], align 4 4959 // CHECK43-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 4960 // CHECK43-NEXT: store i32 [[TMP0]], ptr [[__VLA_EXPR0]], align 4 4961 // CHECK43-NEXT: [[TMP2:%.*]] = load i32, ptr [[N]], align 4 4962 // CHECK43-NEXT: store i32 [[TMP2]], ptr [[DOTCAPTURE_EXPR_]], align 4 4963 // CHECK43-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 4964 // CHECK43-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP3]], 0 4965 // CHECK43-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 4966 // CHECK43-NEXT: [[SUB2:%.*]] = sub nsw i32 [[DIV]], 1 4967 // CHECK43-NEXT: store i32 [[SUB2]], ptr [[DOTCAPTURE_EXPR_1]], align 4 4968 // CHECK43-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 4969 // CHECK43-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 4970 // CHECK43-NEXT: store i32 [[TMP4]], ptr [[DOTOMP_UB]], align 4 4971 // CHECK43-NEXT: store i32 0, ptr [[I]], align 4 4972 // CHECK43-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 4973 // CHECK43-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP5]] 4974 // CHECK43-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 4975 // CHECK43: simd.if.then: 4976 // CHECK43-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 4977 // CHECK43-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4 4978 // CHECK43-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 4979 // CHECK43: omp.inner.for.cond: 4980 // CHECK43-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3:![0-9]+]] 4981 // CHECK43-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP3]] 4982 // CHECK43-NEXT: [[CMP4:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]] 4983 // CHECK43-NEXT: br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 4984 // CHECK43: omp.inner.for.body: 4985 // CHECK43-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 4986 // CHECK43-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1 4987 // CHECK43-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 4988 // CHECK43-NEXT: store i32 [[ADD]], ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP3]] 4989 // CHECK43-NEXT: [[TMP10:%.*]] = load i32, ptr [[I3]], align 4, !llvm.access.group [[ACC_GRP3]] 4990 // CHECK43-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i32 [[TMP10]] 4991 // CHECK43-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP3]] 4992 // CHECK43-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 4993 // CHECK43: omp.body.continue: 4994 // CHECK43-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 4995 // CHECK43: omp.inner.for.inc: 4996 // CHECK43-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 4997 // CHECK43-NEXT: [[ADD5:%.*]] = add nsw i32 [[TMP11]], 1 4998 // CHECK43-NEXT: store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 4999 // CHECK43-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] 5000 // CHECK43: omp.inner.for.end: 5001 // CHECK43-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_]], align 4 5002 // CHECK43-NEXT: [[SUB6:%.*]] = sub nsw i32 [[TMP12]], 0 5003 // CHECK43-NEXT: [[DIV7:%.*]] = sdiv i32 [[SUB6]], 1 5004 // CHECK43-NEXT: [[MUL8:%.*]] = mul nsw i32 [[DIV7]], 1 5005 // CHECK43-NEXT: [[ADD9:%.*]] = add nsw i32 0, [[MUL8]] 5006 // CHECK43-NEXT: store i32 [[ADD9]], ptr [[I3]], align 4 5007 // CHECK43-NEXT: br label [[SIMD_IF_END]] 5008 // CHECK43: simd.if.end: 5009 // CHECK43-NEXT: [[TMP13:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 5010 // CHECK43-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10EEiT_(i32 noundef [[TMP13]]) 5011 // CHECK43-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 5012 // CHECK43-NEXT: [[TMP14:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4 5013 // CHECK43-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP14]]) 5014 // CHECK43-NEXT: [[TMP15:%.*]] = load i32, ptr [[RETVAL]], align 4 5015 // CHECK43-NEXT: ret i32 [[TMP15]] 5016 // 5017 // 5018 // CHECK43-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 5019 // CHECK43-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] comdat { 5020 // CHECK43-NEXT: entry: 5021 // CHECK43-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 5022 // CHECK43-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 5023 // CHECK43-NEXT: [[TE:%.*]] = alloca i32, align 4 5024 // CHECK43-NEXT: [[TH:%.*]] = alloca i32, align 4 5025 // CHECK43-NEXT: [[TMP:%.*]] = alloca i32, align 4 5026 // CHECK43-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 5027 // CHECK43-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 5028 // CHECK43-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 5029 // CHECK43-NEXT: [[I:%.*]] = alloca i32, align 4 5030 // CHECK43-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 5031 // CHECK43-NEXT: store i32 0, ptr [[TE]], align 4 5032 // CHECK43-NEXT: store i32 128, ptr [[TH]], align 4 5033 // CHECK43-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 5034 // CHECK43-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 5035 // CHECK43-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 5036 // CHECK43-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 5037 // CHECK43-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 5038 // CHECK43: omp.inner.for.cond: 5039 // CHECK43-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7:![0-9]+]] 5040 // CHECK43-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP7]] 5041 // CHECK43-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] 5042 // CHECK43-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 5043 // CHECK43: omp.inner.for.body: 5044 // CHECK43-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]] 5045 // CHECK43-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 5046 // CHECK43-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 5047 // CHECK43-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP7]] 5048 // CHECK43-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP7]] 5049 // CHECK43-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[A]], i32 0, i32 [[TMP4]] 5050 // CHECK43-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP7]] 5051 // CHECK43-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 5052 // CHECK43: omp.body.continue: 5053 // CHECK43-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 5054 // CHECK43: omp.inner.for.inc: 5055 // CHECK43-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]] 5056 // CHECK43-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], 1 5057 // CHECK43-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP7]] 5058 // CHECK43-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP8:![0-9]+]] 5059 // CHECK43: omp.inner.for.end: 5060 // CHECK43-NEXT: store i32 10, ptr [[I]], align 4 5061 // CHECK43-NEXT: ret i32 0 5062 // 5063 // 5064 // CHECK45-LABEL: define {{[^@]+}}@main 5065 // CHECK45-SAME: (i32 noundef signext [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 5066 // CHECK45-NEXT: entry: 5067 // CHECK45-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 5068 // CHECK45-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 5069 // CHECK45-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 8 5070 // CHECK45-NEXT: [[N:%.*]] = alloca i32, align 4 5071 // CHECK45-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 8 5072 // CHECK45-NEXT: [[__VLA_EXPR0:%.*]] = alloca i64, align 8 5073 // CHECK45-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 5074 // CHECK45-NEXT: [[TMP:%.*]] = alloca i32, align 4 5075 // CHECK45-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 5076 // CHECK45-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 5077 // CHECK45-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 5078 // CHECK45-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 5079 // CHECK45-NEXT: [[I:%.*]] = alloca i32, align 4 5080 // CHECK45-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 5081 // CHECK45-NEXT: [[I4:%.*]] = alloca i32, align 4 5082 // CHECK45-NEXT: store i32 0, ptr [[RETVAL]], align 4 5083 // CHECK45-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 5084 // CHECK45-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 8 5085 // CHECK45-NEXT: store i32 100, ptr [[N]], align 4 5086 // CHECK45-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 5087 // CHECK45-NEXT: [[TMP1:%.*]] = zext i32 [[TMP0]] to i64 5088 // CHECK45-NEXT: [[TMP2:%.*]] = call ptr @llvm.stacksave.p0() 5089 // CHECK45-NEXT: store ptr [[TMP2]], ptr [[SAVED_STACK]], align 8 5090 // CHECK45-NEXT: [[VLA:%.*]] = alloca i32, i64 [[TMP1]], align 4 5091 // CHECK45-NEXT: store i64 [[TMP1]], ptr [[__VLA_EXPR0]], align 8 5092 // CHECK45-NEXT: [[TMP3:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 5093 // CHECK45-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP3]], 0 5094 // CHECK45-NEXT: [[STOREDV:%.*]] = zext i1 [[TOBOOL]] to i8 5095 // CHECK45-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR_]], align 1 5096 // CHECK45-NEXT: [[TMP4:%.*]] = load i32, ptr [[N]], align 4 5097 // CHECK45-NEXT: store i32 [[TMP4]], ptr [[DOTCAPTURE_EXPR_1]], align 4 5098 // CHECK45-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 5099 // CHECK45-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP5]], 0 5100 // CHECK45-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 5101 // CHECK45-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 5102 // CHECK45-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4 5103 // CHECK45-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 5104 // CHECK45-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 5105 // CHECK45-NEXT: store i32 [[TMP6]], ptr [[DOTOMP_UB]], align 4 5106 // CHECK45-NEXT: store i32 0, ptr [[I]], align 4 5107 // CHECK45-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 5108 // CHECK45-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP7]] 5109 // CHECK45-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 5110 // CHECK45: simd.if.then: 5111 // CHECK45-NEXT: [[TMP8:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 5112 // CHECK45-NEXT: store i32 [[TMP8]], ptr [[DOTOMP_IV]], align 4 5113 // CHECK45-NEXT: [[TMP9:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1 5114 // CHECK45-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP9]] to i1 5115 // CHECK45-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] 5116 // CHECK45: omp_if.then: 5117 // CHECK45-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 5118 // CHECK45: omp.inner.for.cond: 5119 // CHECK45-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2:![0-9]+]] 5120 // CHECK45-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP2]] 5121 // CHECK45-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP10]], [[TMP11]] 5122 // CHECK45-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 5123 // CHECK45: omp.inner.for.body: 5124 // CHECK45-NEXT: [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 5125 // CHECK45-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP12]], 1 5126 // CHECK45-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 5127 // CHECK45-NEXT: store i32 [[ADD]], ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP2]] 5128 // CHECK45-NEXT: [[TMP13:%.*]] = load i32, ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP2]] 5129 // CHECK45-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP13]] to i64 5130 // CHECK45-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i64 [[IDXPROM]] 5131 // CHECK45-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP2]] 5132 // CHECK45-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 5133 // CHECK45: omp.body.continue: 5134 // CHECK45-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 5135 // CHECK45: omp.inner.for.inc: 5136 // CHECK45-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 5137 // CHECK45-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP14]], 1 5138 // CHECK45-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP2]] 5139 // CHECK45-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP3:![0-9]+]] 5140 // CHECK45: omp.inner.for.end: 5141 // CHECK45-NEXT: br label [[OMP_IF_END:%.*]] 5142 // CHECK45: omp_if.else: 5143 // CHECK45-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]] 5144 // CHECK45: omp.inner.for.cond7: 5145 // CHECK45-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 5146 // CHECK45-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 5147 // CHECK45-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]] 5148 // CHECK45-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END17:%.*]] 5149 // CHECK45: omp.inner.for.body9: 5150 // CHECK45-NEXT: [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 5151 // CHECK45-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP17]], 1 5152 // CHECK45-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]] 5153 // CHECK45-NEXT: store i32 [[ADD11]], ptr [[I4]], align 4 5154 // CHECK45-NEXT: [[TMP18:%.*]] = load i32, ptr [[I4]], align 4 5155 // CHECK45-NEXT: [[IDXPROM12:%.*]] = sext i32 [[TMP18]] to i64 5156 // CHECK45-NEXT: [[ARRAYIDX13:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i64 [[IDXPROM12]] 5157 // CHECK45-NEXT: store i32 0, ptr [[ARRAYIDX13]], align 4 5158 // CHECK45-NEXT: br label [[OMP_BODY_CONTINUE14:%.*]] 5159 // CHECK45: omp.body.continue14: 5160 // CHECK45-NEXT: br label [[OMP_INNER_FOR_INC15:%.*]] 5161 // CHECK45: omp.inner.for.inc15: 5162 // CHECK45-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 5163 // CHECK45-NEXT: [[ADD16:%.*]] = add nsw i32 [[TMP19]], 1 5164 // CHECK45-NEXT: store i32 [[ADD16]], ptr [[DOTOMP_IV]], align 4 5165 // CHECK45-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP6:![0-9]+]] 5166 // CHECK45: omp.inner.for.end17: 5167 // CHECK45-NEXT: br label [[OMP_IF_END]] 5168 // CHECK45: omp_if.end: 5169 // CHECK45-NEXT: [[TMP20:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 5170 // CHECK45-NEXT: [[SUB18:%.*]] = sub nsw i32 [[TMP20]], 0 5171 // CHECK45-NEXT: [[DIV19:%.*]] = sdiv i32 [[SUB18]], 1 5172 // CHECK45-NEXT: [[MUL20:%.*]] = mul nsw i32 [[DIV19]], 1 5173 // CHECK45-NEXT: [[ADD21:%.*]] = add nsw i32 0, [[MUL20]] 5174 // CHECK45-NEXT: store i32 [[ADD21]], ptr [[I4]], align 4 5175 // CHECK45-NEXT: br label [[SIMD_IF_END]] 5176 // CHECK45: simd.if.end: 5177 // CHECK45-NEXT: [[TMP21:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 5178 // CHECK45-NEXT: [[CALL:%.*]] = call noundef signext i32 @_Z5tmainIiLi10EEiT_(i32 noundef signext [[TMP21]]) 5179 // CHECK45-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 5180 // CHECK45-NEXT: [[TMP22:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8 5181 // CHECK45-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP22]]) 5182 // CHECK45-NEXT: [[TMP23:%.*]] = load i32, ptr [[RETVAL]], align 4 5183 // CHECK45-NEXT: ret i32 [[TMP23]] 5184 // 5185 // 5186 // CHECK45-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 5187 // CHECK45-SAME: (i32 noundef signext [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] comdat { 5188 // CHECK45-NEXT: entry: 5189 // CHECK45-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 5190 // CHECK45-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 5191 // CHECK45-NEXT: [[TE:%.*]] = alloca i32, align 4 5192 // CHECK45-NEXT: [[TH:%.*]] = alloca i32, align 4 5193 // CHECK45-NEXT: [[TMP:%.*]] = alloca i32, align 4 5194 // CHECK45-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 5195 // CHECK45-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 5196 // CHECK45-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 5197 // CHECK45-NEXT: [[I:%.*]] = alloca i32, align 4 5198 // CHECK45-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 5199 // CHECK45-NEXT: store i32 0, ptr [[TE]], align 4 5200 // CHECK45-NEXT: store i32 128, ptr [[TH]], align 4 5201 // CHECK45-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 5202 // CHECK45-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 5203 // CHECK45-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 5204 // CHECK45-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 5205 // CHECK45-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 5206 // CHECK45: omp.inner.for.cond: 5207 // CHECK45-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8:![0-9]+]] 5208 // CHECK45-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP8]] 5209 // CHECK45-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] 5210 // CHECK45-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 5211 // CHECK45: omp.inner.for.body: 5212 // CHECK45-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8]] 5213 // CHECK45-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 5214 // CHECK45-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 5215 // CHECK45-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP8]] 5216 // CHECK45-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP8]] 5217 // CHECK45-NEXT: [[IDXPROM:%.*]] = sext i32 [[TMP4]] to i64 5218 // CHECK45-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[A]], i64 0, i64 [[IDXPROM]] 5219 // CHECK45-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP8]] 5220 // CHECK45-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 5221 // CHECK45: omp.body.continue: 5222 // CHECK45-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 5223 // CHECK45: omp.inner.for.inc: 5224 // CHECK45-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8]] 5225 // CHECK45-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], 1 5226 // CHECK45-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP8]] 5227 // CHECK45-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP9:![0-9]+]] 5228 // CHECK45: omp.inner.for.end: 5229 // CHECK45-NEXT: store i32 10, ptr [[I]], align 4 5230 // CHECK45-NEXT: ret i32 0 5231 // 5232 // 5233 // CHECK47-LABEL: define {{[^@]+}}@main 5234 // CHECK47-SAME: (i32 noundef [[ARGC:%.*]], ptr noundef [[ARGV:%.*]]) #[[ATTR0:[0-9]+]] { 5235 // CHECK47-NEXT: entry: 5236 // CHECK47-NEXT: [[RETVAL:%.*]] = alloca i32, align 4 5237 // CHECK47-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 5238 // CHECK47-NEXT: [[ARGV_ADDR:%.*]] = alloca ptr, align 4 5239 // CHECK47-NEXT: [[N:%.*]] = alloca i32, align 4 5240 // CHECK47-NEXT: [[SAVED_STACK:%.*]] = alloca ptr, align 4 5241 // CHECK47-NEXT: [[__VLA_EXPR0:%.*]] = alloca i32, align 4 5242 // CHECK47-NEXT: [[DOTCAPTURE_EXPR_:%.*]] = alloca i8, align 1 5243 // CHECK47-NEXT: [[TMP:%.*]] = alloca i32, align 4 5244 // CHECK47-NEXT: [[DOTCAPTURE_EXPR_1:%.*]] = alloca i32, align 4 5245 // CHECK47-NEXT: [[DOTCAPTURE_EXPR_2:%.*]] = alloca i32, align 4 5246 // CHECK47-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 5247 // CHECK47-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 5248 // CHECK47-NEXT: [[I:%.*]] = alloca i32, align 4 5249 // CHECK47-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 5250 // CHECK47-NEXT: [[I4:%.*]] = alloca i32, align 4 5251 // CHECK47-NEXT: store i32 0, ptr [[RETVAL]], align 4 5252 // CHECK47-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 5253 // CHECK47-NEXT: store ptr [[ARGV]], ptr [[ARGV_ADDR]], align 4 5254 // CHECK47-NEXT: store i32 100, ptr [[N]], align 4 5255 // CHECK47-NEXT: [[TMP0:%.*]] = load i32, ptr [[N]], align 4 5256 // CHECK47-NEXT: [[TMP1:%.*]] = call ptr @llvm.stacksave.p0() 5257 // CHECK47-NEXT: store ptr [[TMP1]], ptr [[SAVED_STACK]], align 4 5258 // CHECK47-NEXT: [[VLA:%.*]] = alloca i32, i32 [[TMP0]], align 4 5259 // CHECK47-NEXT: store i32 [[TMP0]], ptr [[__VLA_EXPR0]], align 4 5260 // CHECK47-NEXT: [[TMP2:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 5261 // CHECK47-NEXT: [[TOBOOL:%.*]] = icmp ne i32 [[TMP2]], 0 5262 // CHECK47-NEXT: [[STOREDV:%.*]] = zext i1 [[TOBOOL]] to i8 5263 // CHECK47-NEXT: store i8 [[STOREDV]], ptr [[DOTCAPTURE_EXPR_]], align 1 5264 // CHECK47-NEXT: [[TMP3:%.*]] = load i32, ptr [[N]], align 4 5265 // CHECK47-NEXT: store i32 [[TMP3]], ptr [[DOTCAPTURE_EXPR_1]], align 4 5266 // CHECK47-NEXT: [[TMP4:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 5267 // CHECK47-NEXT: [[SUB:%.*]] = sub nsw i32 [[TMP4]], 0 5268 // CHECK47-NEXT: [[DIV:%.*]] = sdiv i32 [[SUB]], 1 5269 // CHECK47-NEXT: [[SUB3:%.*]] = sub nsw i32 [[DIV]], 1 5270 // CHECK47-NEXT: store i32 [[SUB3]], ptr [[DOTCAPTURE_EXPR_2]], align 4 5271 // CHECK47-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 5272 // CHECK47-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_2]], align 4 5273 // CHECK47-NEXT: store i32 [[TMP5]], ptr [[DOTOMP_UB]], align 4 5274 // CHECK47-NEXT: store i32 0, ptr [[I]], align 4 5275 // CHECK47-NEXT: [[TMP6:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 5276 // CHECK47-NEXT: [[CMP:%.*]] = icmp slt i32 0, [[TMP6]] 5277 // CHECK47-NEXT: br i1 [[CMP]], label [[SIMD_IF_THEN:%.*]], label [[SIMD_IF_END:%.*]] 5278 // CHECK47: simd.if.then: 5279 // CHECK47-NEXT: [[TMP7:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 5280 // CHECK47-NEXT: store i32 [[TMP7]], ptr [[DOTOMP_IV]], align 4 5281 // CHECK47-NEXT: [[TMP8:%.*]] = load i8, ptr [[DOTCAPTURE_EXPR_]], align 1 5282 // CHECK47-NEXT: [[LOADEDV:%.*]] = trunc i8 [[TMP8]] to i1 5283 // CHECK47-NEXT: br i1 [[LOADEDV]], label [[OMP_IF_THEN:%.*]], label [[OMP_IF_ELSE:%.*]] 5284 // CHECK47: omp_if.then: 5285 // CHECK47-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 5286 // CHECK47: omp.inner.for.cond: 5287 // CHECK47-NEXT: [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3:![0-9]+]] 5288 // CHECK47-NEXT: [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP3]] 5289 // CHECK47-NEXT: [[CMP5:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]] 5290 // CHECK47-NEXT: br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 5291 // CHECK47: omp.inner.for.body: 5292 // CHECK47-NEXT: [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 5293 // CHECK47-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1 5294 // CHECK47-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 5295 // CHECK47-NEXT: store i32 [[ADD]], ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP3]] 5296 // CHECK47-NEXT: [[TMP12:%.*]] = load i32, ptr [[I4]], align 4, !llvm.access.group [[ACC_GRP3]] 5297 // CHECK47-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i32 [[TMP12]] 5298 // CHECK47-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP3]] 5299 // CHECK47-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 5300 // CHECK47: omp.body.continue: 5301 // CHECK47-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 5302 // CHECK47: omp.inner.for.inc: 5303 // CHECK47-NEXT: [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 5304 // CHECK47-NEXT: [[ADD6:%.*]] = add nsw i32 [[TMP13]], 1 5305 // CHECK47-NEXT: store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP3]] 5306 // CHECK47-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP4:![0-9]+]] 5307 // CHECK47: omp.inner.for.end: 5308 // CHECK47-NEXT: br label [[OMP_IF_END:%.*]] 5309 // CHECK47: omp_if.else: 5310 // CHECK47-NEXT: br label [[OMP_INNER_FOR_COND7:%.*]] 5311 // CHECK47: omp.inner.for.cond7: 5312 // CHECK47-NEXT: [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 5313 // CHECK47-NEXT: [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4 5314 // CHECK47-NEXT: [[CMP8:%.*]] = icmp sle i32 [[TMP14]], [[TMP15]] 5315 // CHECK47-NEXT: br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY9:%.*]], label [[OMP_INNER_FOR_END16:%.*]] 5316 // CHECK47: omp.inner.for.body9: 5317 // CHECK47-NEXT: [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 5318 // CHECK47-NEXT: [[MUL10:%.*]] = mul nsw i32 [[TMP16]], 1 5319 // CHECK47-NEXT: [[ADD11:%.*]] = add nsw i32 0, [[MUL10]] 5320 // CHECK47-NEXT: store i32 [[ADD11]], ptr [[I4]], align 4 5321 // CHECK47-NEXT: [[TMP17:%.*]] = load i32, ptr [[I4]], align 4 5322 // CHECK47-NEXT: [[ARRAYIDX12:%.*]] = getelementptr inbounds i32, ptr [[VLA]], i32 [[TMP17]] 5323 // CHECK47-NEXT: store i32 0, ptr [[ARRAYIDX12]], align 4 5324 // CHECK47-NEXT: br label [[OMP_BODY_CONTINUE13:%.*]] 5325 // CHECK47: omp.body.continue13: 5326 // CHECK47-NEXT: br label [[OMP_INNER_FOR_INC14:%.*]] 5327 // CHECK47: omp.inner.for.inc14: 5328 // CHECK47-NEXT: [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4 5329 // CHECK47-NEXT: [[ADD15:%.*]] = add nsw i32 [[TMP18]], 1 5330 // CHECK47-NEXT: store i32 [[ADD15]], ptr [[DOTOMP_IV]], align 4 5331 // CHECK47-NEXT: br label [[OMP_INNER_FOR_COND7]], !llvm.loop [[LOOP7:![0-9]+]] 5332 // CHECK47: omp.inner.for.end16: 5333 // CHECK47-NEXT: br label [[OMP_IF_END]] 5334 // CHECK47: omp_if.end: 5335 // CHECK47-NEXT: [[TMP19:%.*]] = load i32, ptr [[DOTCAPTURE_EXPR_1]], align 4 5336 // CHECK47-NEXT: [[SUB17:%.*]] = sub nsw i32 [[TMP19]], 0 5337 // CHECK47-NEXT: [[DIV18:%.*]] = sdiv i32 [[SUB17]], 1 5338 // CHECK47-NEXT: [[MUL19:%.*]] = mul nsw i32 [[DIV18]], 1 5339 // CHECK47-NEXT: [[ADD20:%.*]] = add nsw i32 0, [[MUL19]] 5340 // CHECK47-NEXT: store i32 [[ADD20]], ptr [[I4]], align 4 5341 // CHECK47-NEXT: br label [[SIMD_IF_END]] 5342 // CHECK47: simd.if.end: 5343 // CHECK47-NEXT: [[TMP20:%.*]] = load i32, ptr [[ARGC_ADDR]], align 4 5344 // CHECK47-NEXT: [[CALL:%.*]] = call noundef i32 @_Z5tmainIiLi10EEiT_(i32 noundef [[TMP20]]) 5345 // CHECK47-NEXT: store i32 [[CALL]], ptr [[RETVAL]], align 4 5346 // CHECK47-NEXT: [[TMP21:%.*]] = load ptr, ptr [[SAVED_STACK]], align 4 5347 // CHECK47-NEXT: call void @llvm.stackrestore.p0(ptr [[TMP21]]) 5348 // CHECK47-NEXT: [[TMP22:%.*]] = load i32, ptr [[RETVAL]], align 4 5349 // CHECK47-NEXT: ret i32 [[TMP22]] 5350 // 5351 // 5352 // CHECK47-LABEL: define {{[^@]+}}@_Z5tmainIiLi10EEiT_ 5353 // CHECK47-SAME: (i32 noundef [[ARGC:%.*]]) #[[ATTR2:[0-9]+]] comdat { 5354 // CHECK47-NEXT: entry: 5355 // CHECK47-NEXT: [[ARGC_ADDR:%.*]] = alloca i32, align 4 5356 // CHECK47-NEXT: [[A:%.*]] = alloca [10 x i32], align 4 5357 // CHECK47-NEXT: [[TE:%.*]] = alloca i32, align 4 5358 // CHECK47-NEXT: [[TH:%.*]] = alloca i32, align 4 5359 // CHECK47-NEXT: [[TMP:%.*]] = alloca i32, align 4 5360 // CHECK47-NEXT: [[DOTOMP_LB:%.*]] = alloca i32, align 4 5361 // CHECK47-NEXT: [[DOTOMP_UB:%.*]] = alloca i32, align 4 5362 // CHECK47-NEXT: [[DOTOMP_IV:%.*]] = alloca i32, align 4 5363 // CHECK47-NEXT: [[I:%.*]] = alloca i32, align 4 5364 // CHECK47-NEXT: store i32 [[ARGC]], ptr [[ARGC_ADDR]], align 4 5365 // CHECK47-NEXT: store i32 0, ptr [[TE]], align 4 5366 // CHECK47-NEXT: store i32 128, ptr [[TH]], align 4 5367 // CHECK47-NEXT: store i32 0, ptr [[DOTOMP_LB]], align 4 5368 // CHECK47-NEXT: store i32 9, ptr [[DOTOMP_UB]], align 4 5369 // CHECK47-NEXT: [[TMP0:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4 5370 // CHECK47-NEXT: store i32 [[TMP0]], ptr [[DOTOMP_IV]], align 4 5371 // CHECK47-NEXT: br label [[OMP_INNER_FOR_COND:%.*]] 5372 // CHECK47: omp.inner.for.cond: 5373 // CHECK47-NEXT: [[TMP1:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9:![0-9]+]] 5374 // CHECK47-NEXT: [[TMP2:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4, !llvm.access.group [[ACC_GRP9]] 5375 // CHECK47-NEXT: [[CMP:%.*]] = icmp sle i32 [[TMP1]], [[TMP2]] 5376 // CHECK47-NEXT: br i1 [[CMP]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]] 5377 // CHECK47: omp.inner.for.body: 5378 // CHECK47-NEXT: [[TMP3:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9]] 5379 // CHECK47-NEXT: [[MUL:%.*]] = mul nsw i32 [[TMP3]], 1 5380 // CHECK47-NEXT: [[ADD:%.*]] = add nsw i32 0, [[MUL]] 5381 // CHECK47-NEXT: store i32 [[ADD]], ptr [[I]], align 4, !llvm.access.group [[ACC_GRP9]] 5382 // CHECK47-NEXT: [[TMP4:%.*]] = load i32, ptr [[I]], align 4, !llvm.access.group [[ACC_GRP9]] 5383 // CHECK47-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds [10 x i32], ptr [[A]], i32 0, i32 [[TMP4]] 5384 // CHECK47-NEXT: store i32 0, ptr [[ARRAYIDX]], align 4, !llvm.access.group [[ACC_GRP9]] 5385 // CHECK47-NEXT: br label [[OMP_BODY_CONTINUE:%.*]] 5386 // CHECK47: omp.body.continue: 5387 // CHECK47-NEXT: br label [[OMP_INNER_FOR_INC:%.*]] 5388 // CHECK47: omp.inner.for.inc: 5389 // CHECK47-NEXT: [[TMP5:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9]] 5390 // CHECK47-NEXT: [[ADD1:%.*]] = add nsw i32 [[TMP5]], 1 5391 // CHECK47-NEXT: store i32 [[ADD1]], ptr [[DOTOMP_IV]], align 4, !llvm.access.group [[ACC_GRP9]] 5392 // CHECK47-NEXT: br label [[OMP_INNER_FOR_COND]], !llvm.loop [[LOOP10:![0-9]+]] 5393 // CHECK47: omp.inner.for.end: 5394 // CHECK47-NEXT: store i32 10, ptr [[I]], align 4 5395 // CHECK47-NEXT: ret i32 0 5396 // 5397